A Comment on IDA’s Debt Sustainability Framework

In a recent note, we asked whether IDA, the arm of the World Bank that provides concessional lending to the poorest countries, has the right approach for determining the balance between grants and loans which its recipients receive. We noted that IDA concessionality depends on the level of recipients’ debt distress. This raises a concern over potentially perverse incentives.

What is the current system for determining concessionality?

IDA’s Grant Allocation Framework (GAF) (also referred to as the “traffic light system”) makes debt distress the sole determinant of the level of concessionality of allocations. Countries at high risk of, or assessed currently to be in, debt distress (red light) receive all their performance-based IDA allocation in the form of grants; medium-risk countries (yellow light) receive 50 percent as grants and 50 percent as concessional loans; and low-risk countries (green light) receive their allocation solely as a concessional loan.

Why is this problematic?

The risk is that the GAF creates a moral hazard problem. Poor performance in debt sustainability is rewarded with more concessional country allocations. Policy actions that increase the risk of debt distress (which might include running large fiscal deficits over a prolonged period and taking on substantial amounts of high-cost debt) get “rewarded” by an increase in the concessionality of IDA future flows. Actions to reduce the risk of debt distress—fiscal discipline and the avoidance of non-concessional debt—are “punished” with a reduction in the level of concessionality of future flows. A key feature is that the GAF doesn’t ask how a country came to be at risk of debt distress. A high-risk fragile country that has experienced a major natural disaster is treated the same way as a high-risk country that has run large fiscal deficits for years and has financed those deficits with international bonds on non-concessional terms.

How can this be addressed?

The Inter-American Development Bank (IDB) adopted a new approach in 2021 to address this concern in its operations. They added three additional variables to their system for determining the financing blend and concessionality:

  1. Relative poverty: Concessionality declines as GNI per capita increases. A country with a low GNI per capita receives more concessionality points, while countries with a GNI per capita above a threshold of $2,834 receive no concessionality points.
  2. Index of vulnerability: A composite index using six existing publicly available indices covering different facets of vulnerability (small population size, island or landlocked status, remoteness from world markets, exposure to natural hazards, vulnerability to climate change, and fragility). More vulnerable countries, as reflected by higher scores on the vulnerability index, receive more concessionality points.
  3. An index of non-concessional borrowing: More non-concessional borrowing reduces a country’s concessionality score. The extent of non-concessional borrowing is measured as the sum of public and publicly guaranteed external debt from private creditors and non-concessional bilateral creditors, divided by the country’s GNI.

Relative poverty, vulnerability, and debt distress are equally weighted in the IDB formula, and the concessionality level is then reduced based on the level of non-concessional borrowing.

The impact of the IDB proposal is illustrated by the comparison between Maldives and Uganda below:

  Characteristics IDA-FY2020 IADB MODEL Characteristics IDA-FY2020 IADB MODEL
Grant element   80% 42%   53% 62%
GNI per capita US$9,310 Gap country status No concessionality points from income per capita US$620 IDA-only status Low income per capita contributes strongly to concessionality points
Small island / landlocked Small island Small island status significantly boosts concessionality Small population and island status boosts vulnerability score Landlocked No impact Landlocked status contributes to vulnerability score
Fragility Ninth least fragile IDA country according to Fragile States Index Does not affect concessionality Low fragility counteracts vulnerability points from small population and island status 24th most fragile IDA country according to Fragile States Index No impact Adds to concessionality level (through vulnerability score)
High risk of debt distress Dominant characteristic High risk: maximum points from this consideration (but is offset by other considerations) Low risk of debt distress Key characteristic that substantially reduces level of concessionality Low risk: minimum points from this consideration (but is offset by other considerations)
Non-concessionality borrowing Stock of non-concessional borrowing = 33.9% of GDP Application of non-concessional borrowing policy remedies reduces grant element by 20 percentage points Very high level of non-concessional borrowing completely offsets risk of debt distress rating Stock of non-concessional borrowing = 8.1% of GDP No impact Moderate level of non-concessional borrowing reduces grant element by several percentage points

To summarise, Maldives has a relatively high GNI per capita, substantial non-concessional borrowing, and low fragility. It is at high risk of debt distress. In the 2020 financial year, Maldives received half of its IDA allocation in grants and half its allocation in concessional loans with a grant element of 60 percent. Overall, its allocation had a very high grant element of 80 percent. By contrast, under the IDB model it would have received an allocation with a grant element of only 42 percent. On the other hand, Uganda, with a relatively low GNI per capita and a moderate level of vulnerability, is at low risk of debt distress. It received its 2020 IDA allocation entirely as concessional loans with a grant element of 53 percent. Had the IDB model been applied, Uganda would have received an allocation with a grant element of 62 percent.

What difference would adopting the IDB system more widely across the multilateral development banks make?

For some countries, possibly not much. Where, for example, debt distress is highly correlated with fragility and low per capita GNI, as often happens, the current GAF and the IDB system may produce similar results.

For others, however, the difference might be more significant. Countries like Malawi, Nepal, Rwanda, and Tanzania (as well as Uganda), which are relatively poor and fragile but responsible in limiting their non-concessional borrowing, are penalised with a lower level of concessionality than some others receive.

At minimum there is an optical issue. If the MDBs want to signal the desirability of responsible borrowing, adopting the IDB approach would help. The combination of providing higher concessionality for those with higher risk of debt distress and subtracting concessionality points with increasing non-concessional borrowing, allows a distinction to be made between countries at high risk of debt distress for structural reasons or due to an exogenous shock, and those at high risk of debt distress as a result of their level of non-concessional borrowing.

Note: We are extremely grateful to Dougal Martin of the Inter-American Development Bank for detailed and helpful comments in the preparation of this.


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.