Steve Cutts: Credit Where Credit's Due, February 2022
Updated: Oct 23
About this paper
Why does the way aid loans are counted matter? In this paper, Stephen J Cutts examines how the OECD Development Assistance Committee (DAC) measures Official Development Assistance (ODA) in concessional loans and debt relief. He argues that the grant equivalent methodology overstates donor effort and creates incentives to provide loans rather than grants.
First sent to the DAC Chair in February 2022, this paper sets out five methodological flaws, six recommendations, and their implications for development assistance and climate finance. The full text below includes the author’s note, all nine annexes, worked examples, and source notes.
Editorial note: This is a web transcription of the paper dated 18 February 2022. References to current rates, country classifications, policy reviews and climate finance commitments belong to that historical context. The original wording and figures are retained; this is not an update of the analysis to 2026.
Giving Credit Where Credit’s Due: The Need to Address Flaws in the Calculation of ODA in Loans
Paper by Stephen J Cutts, 18 February 2022
Background
This paper sets out some serious shortcomings in the methodology that the Development Assistance Committee (DAC) uses for calculating Official Development Assistance in concessional loans and debt relief. These flaws have serious impacts in undermining the credibility of the statistics produced by the OECD/DAC and misleading the public by massively over-stating the ODA in concessional loans. The flawed methodology also incentivises governments to provide ODA loans rather than grants. This seems to be having an especially pernicious impact around climate finance, where developing countries are being saddled with unsustainable debts to deal with the costs brought about by climate change that is not of their making.
These issues call into serious question the legitimacy of the DAC – an OECD committee comprising only donors with incentives to exaggerate the ODA they provide – continuing to have stewardship of this issue.
I. Calculating ODA in Loans Before 2014/18
I first became involved in this issue some 20 years ago, back in 2002 when, as Head of the Office of an OECD Deputy Secretary-General, I wrote to the DAC Secretariat drawing attention to serious flaws in the previous methodology for calculating the ODA element of concessional loans, that produced nonsensical results. Annex 1 sets out the main flaws in the methodology that I, along with others, identified at that time.
Noting that the 10% discount rate, even then, was far too high given prevailing interest rates and long-term bond yields, we proposed a new system whereby only the grant element of a loan would count as ODA, and that lower and differentiated discount rates should be used in the calculation of that grant element. The proposals were strongly resisted by the DAC Secretariat at the time, vigorously defending the prevailing method as robust, and it was only ditched many years later.
II. DAC Reform, 2014 (Effective 2018)
Against this background, it was a pleasant surprise to learn that DAC members decided, at their 2014 High-Level Meeting, to adopt a new system for measuring ODA, now counting only the “grant element” of loans as ODA and using lower discount rates (both as proposed in 2002), and that this new system would be used as the basis for calculating DAC members’ progress towards the well-known 0.7% of GNI target for ODA from 2018 onwards.
However, upon further analysis, there are serious flaws in the new methodology and, ironically, although the DAC claimed at the time of announcing the changes that the “modernised” way of counting ODA provided stronger incentives to use grants[1], it clearly does the opposite. Under the previous system, over its life, every loan yielded a net ODA of zero, regardless of how soft its terms were. While a donor could claim the full-face value of a loan as ODA on disbursement, every repayment of principal was counted as negative ODA, thereby eating away at the initial ODA credit, until it was fully cancelled out. (Interest rate payments were simply ignored as to do otherwise would have resulted in any interest-bearing loan, no matter how low the rate, counting as negative ODA.) This methodology was clearly wrong, but it did incentivise the provision of grants over loans, at least over the longer term.
While counting only the grant element of a loan as ODA is fundamentally the right approach, it is vital under such a system that the discount rate used to calculate that grant element is not set too high. If it is, the financial incentive for donors in terms of measured ODA to provide loans rather than grants is effectively “baked in” for the duration of a loan. Yet this is exactly what the DAC have done.
III. Setting the Discount Rate
Since 2018, the DAC has recorded the grant equivalent of loans as ODA, which in turn the DAC defines as a “measure of donor effort”[2]. Accordingly, the ODA counted for a loan (or the grant equivalent) should be the amount that the loan is expected to cost the donor in financial terms. This means that the discount rate needs to be set at the same level as the interest rate for a loan that the donor would need to cover its costs, and no more. Using such a discount rate, any shortfall against the face value of the loan in a net present value calculation of future repayments of principal and interest represents the net cost to the government for extending the loan (the donor effort), and any surplus from such a calculation shows the net financial benefit to the donor.
The costs a donor bears in extending a loan comprise two elements:
The cost of funds for the donor. This is what it would cost the donor to borrow funds on its own behalf.
The cost of any losses that might occur due to the borrower defaulting on repayments or due to debt relief being needed, whether resulting in the rescheduling of repayments or the forgiveness of part or all of the debt. This is the credit risk.
IV. Flaws in the Grant Equivalent Methodology
Flaw 1: The Use of the IMF’s 5% Discount Rate as the “base factor” of the DAC’s Discount Rates is Far Higher than Donors’ Cost of Funds
The first cost that needs to be covered in the discount rate is the donor’s cost of funds.
In this connection, the DAC has chosen a “base factor of 5% for all currencies and loan periods”. The DAC justified this decision by noting that it is consistent with the discount rates that the IMF has used since 2014 for calculating the grant element in its loans[3].
It is true that in 2013, the IMF decided to move away from using differential discount rates for different currencies and instead agreed to use a uniform discount rate of 5%, which was linked to the historic 10-year average of the US Dollar Commercial Interest Reference Rates (CIRR)[4].
The IMF explained its organization-specific reasons for doing so in a technical note[5]. These included
the previous system of using multiple CIRRs for different currencies was complex;
the reduction in discount rates had pushed up the value of LIC debt and was thereby narrowing the assessed borrowing space of these countries;
interest rates across different currencies were converging; and
the IMF was increasingly providing loans in currencies for which CIRRs were unavailable.
None of these justify the decision of the DAC to follow the IMF practice. The DAC is supposed to be measuring the bilateral donor effort in extending loans; it should therefore use discount rates that reflect the funding costs of individual lending governments (which are far below 5% for almost all currencies) and not a figure based on outdate lending rates in one currency. (Even if a global benchmark were to be approved, it should be set at a much lower figure: the US CIRR between 2012 and now has consistently been much lower than 5% and is currently less than 3%.)
The cost of finance for an individual government borrowing in its own domestic currency is determined by the current yield on its own government bonds at the relevant maturity. These vary between different countries and whether they are borrowing in domestic or foreign currencies. This explains why OECD Governments have since 1991 agreed to use Differentiated Discount Rates (DDRs) for each individual currency for the purpose of calculating the concessionality level of tied aid loans. These DDRs are updated and published by the OECD on an annual basis.
The DDRs are a much closer approximation of the costs of borrowing for donors. The DDRs are calculated from a 6-month average of Commercial Interest Reference Rates (CIRRs) for each currency (also compiled and published by the OECD), and these CIRRs are, in turn, based on government bond yields. A margin is added, depending on the maturity (duration) of the loan. In fact, CIRRs (and the DDRs) already include an additional margin of 100 basis points (as they are intended to “represent final commercial lending rates in the domestic market of the currency concerned…and should closely correspond to the rate for first class domestic…[and] foreign borrowers.”) Accordingly, even DDRs are higher than the true cost of government finance[6]. Nevertheless, as they are accepted by OECD governments as proxies for calculating the concessionality level of tied aid loans, DDRs have been used in the calculations for this paper as a proxy for the cost of funds.
The new DDRs have just been published: https://www.oecd.org/trade/topics/export-credits/documents/ddr-tad.pdf (also at Annex 8). As can be seen, these discount rates vary significantly but they are, for the most part, far below the 5% base rate currently being used by the DAC.
It is particularly noticeable that the DDRs for the currencies of donors who currently provide a substantial part of their ODA through loans are much lower than 5%. For example, the Japanese Yen DDR ranges between 1.7% and 2.2% (depending on maturity) and the DDR for the Euro (as used by France, Germany, etc.) ranges from 1.3% to 1.8%. Japan, France and Germany are amongst those countries that provide a substantial proportion of their development assistance through loans rather than grants. This is perhaps unsurprising as these countries are highly incentivised to extend loans as they benefit considerably (i.e. receive considerable “unearned credit”) from the difference between the much higher discount rate used by the DAC and their own relatively low cost of funds.
By choosing to align itself with the IMF discount rate of 5% instead, for example, of the OECD’s own DDRs used to calculate the concessionality level of tied aid loans, the DAC is crediting donors who provide concessional loans with ODA far exceeding the costs and effort those donors incur in extending those loans to developing countries. These DDRs, ideally without the 100 basis points margin in the CIRRs, would be a far more legitimate base factor in the discount rate for determining donor effort than the flat 5% currently used by the DAC.
Recommendation 1: The base factor of 5% in the discount rate is massively inflating the ODA credited for major donors. Discount rates should be differentiated by currency in a way that reflects more accurately the true funding costs of donor governments
Flaw 2: It is unjustifiable to include an adjustment in the discount rate for credit risk and then also to count the costs of debt relief when it occurs as additional ODA
The second cost the donor faces in extending the loan is the financial shortfall the donor may suffer due to the borrower not repaying the loan in full, together with agreed interest. This is the credit risk. A commercial lender needs to include a risk premium, payable either as a fee or as an addition to the interest rate spread, to build sufficient reserves to meet the costs of expected future credit risk losses. Government donors have two options: make a risk-related adjustment to the discount rate upfront so that, effectively, donors are counting additional ODA that should equate with the costs of future losses (if the risk pricing is correct); or ignore the credit risk in the discount rate and simply claim additional ODA equal to the costs borne by a donor when debt relief is needed and costs are incurred. (It is noteworthy that the DDRs discussed above include no adjustment for the credit risk of the borrower.) Annex 3 explores these two options as well as a “hybrid” option.
The DAC has decided to adjust the discount rate to account for credit risk. Accordingly, to the base rate of 5% (that is intended to reflect the donor’s own cost of finance), the DAC adds a margin of 1% for loans to multilateral institutions (other than global institutions and multilateral banks for which no margin is added as their credit risk is deemed negligible) as well as for bilateral loans to Upper Middle Income Countries (UMICs); 2% for bilateral loans to Lower Middle Income Countries (LMICs) and 4% for bilateral loans to Least Developed Countries (LDCs) and other Low Income Countries (LICs). It is explained, and widely understood, that these “adjustment factors” aim to cover the credit risk, including delayed repayment or the need for debt forgiveness.
On the face of it, an additional charge for credit risk is reasonable – after all, there are additional donor costs and hence effort when a loan is rescheduled or forgiven. Moreover, as shown by the calculations at Annex 9, the adjustments made to the discount rate are probably more than sufficient to account for the credit risk (they are generally higher than the credit risk premium rates for sovereign borrowers charged by export credit agencies, which OECD countries consider adequate to cover long term operating costs and losses - see text under Flaw 3).
However, the inclusion within ODA of what is effectively a risk premium needs to be placed in the context of decisions the DAC made in 2020 concerning the ODA treatment of debt rescheduling/forgiveness, and which are explained in a note “Reporting on Debt Relief in the Grant Equivalent System”[7]. The preamble in the Narrative in section 1.2 in this note recalls that in 2014, the DAC “had agreed that the cost of risk should…not be double counted”. However, the decision of the DAC to allow the additional “grant element” incurred for the costs of rescheduling or debt forgiveness does just that.
Adding the costs of debt relief as additional ODA negates the rationale for adding a margin upfront to the discount rate. This is effectively double counting as ODA is being added in the form of risk premiums charged to reflect the credit risk of debt relief and then again for the costs associated with the same debt relief: ODA is effectively being counted twice for the same risk[8]. Put another way, the DAC’s decision to allow the costs of debt relief when it occurs to be added to ODA means that there is no donor effort that is not already being recorded as ODA that would warrant any additional cost that needs to be “credited” upfront as ODA. If additional ODA is going to be credited for the costs of any default, restructuring or loan forgiveness, the need for any credit risk in the discount rate is redundant.
The only limitation to additional ODA counted is that a ceiling is applied so that the new ODA for the additional costs resulting from the debt relief when added to the ODA originally recorded for the loan cannot exceed the face value of the loan[9]. In other words, the total ODA for a loan can never exceed the ODA value of a grant of an amount equivalent to the face value of the loan, regardless of the level or timing of any debt relief. However, this caveat neither negates nor significantly mitigates the double-counting problem.
Annex 2 to this paper offers detailed rebuttal of the DAC’s justification and explanations for its decision to allow debt relief to count as additional ODA while maintaining the risk adjustment in the discount rate.
Recommendation 2: The double-counting of risk by scoring ODA both for assuming the risk of non-repayment up-front and again for bearing the cost of actual non-repayments should be stopped.
Flaw 3: The assumption that credit risk is closely correlated with World Bank Income Classification is flawed
Discussion of this issue needs to be heavily caveated because, as explained in the preceding paragraphs, there is no justification for the DAC adding any spread for credit risk to the discount rate for calculating ODA, when it has separately decided that the costs of any debt relief, whether rescheduling or forgiveness, will count as additional ODA.
Nevertheless, it is important to examine the assumptions the DAC makes about credit risk. Firstly, because the decision to allow the costs of debt relief as additional ODA could be reversed; and secondly because even if that decision stands and the credit risk adjustment in the discount rate is removed, there is a risk of under-counting ODA that needs to be addressed. This might occur for a loan where the interest rate is insufficient to cover the credit risk in full, yet the loan does not qualify as an ODA loan due to its low or even negative grant element. Such loans would need to be tracked and the uncovered share of the credit risk assessed, as this share would need to be applied to the costs of any debt relief[10] and counted as additional ODA.
With respect to its assessment of credit risk, in its Grant Equivalent system, as noted above, the DAC adds an adjustment factor for credit risk of 1% for UMICs, 2% for LMICs and 4% for LDCs and other LICs. This therefore assumes that countries in these economic groups defined by income[11] have a broadly similar credit risk. This assumption is wrong.
While the DAC recognizes the need to monitor experience over time “to reflect evidence gathered on risk”[12] it has seemingly overlooked data from the major credit rating agencies (such as Fitch, Moodys and Standard & Poors). Perhaps more pertinently, it has not taken into account the results of the extensive work on country credit risk undertaken by the Participants to the Export Credit Arrangement supported by the OECD’s own Export Credits Division. The results of this work, which are regularly updated, are highly relevant for the DAC’s methodology. Accordingly, it is useful to set out the process and rationale in brief.
The Subsidies Code of the World Trade Organisation (WTO) requires that premiums are collected when any Government provides official export credit support. These premiums must be risk-related and sufficient to cover the risk of future losses[13].
Given this requirement, Country Risk Experts (CREs) representing the Participants to the Arrangement on Officially Support Export Credits developed a Model some 25 years ago for assessing country risk for virtually all developing countries, and a process for adjusting the results of that model as deemed appropriate by a group of highly experienced country risk experts from the OECD Governments who comprise the Participants to the Export Credit Arrangement[14].
A full explanation of the methodology for assessing country risk and the placement of countries into seven different risk categories is set out on the OECD website[15] and summarized in Annex 4.
The resultant country risk classifications are made public by the OECD Secretariat; the latest listing is here: https://www.oecd.org/trade/topics/export-credits/documents/cre-crc-current-english.pdf. They have been reproduced, by country category, at Annex 5.
The outcomes of the country credit risk methodology show clearly that the DAC approach of equating the different income categories of countries with their credit risk is flawed. The correlation is not strong. While it is true that all LICs and LDCs fall within Categories 6 and 7 (the highest credit risk), UMICs and LMICs are scattered across the different risk categories.
Specifically, UMICs are found in Groups 2 through 7 and LMICs are found in Groups 3 through 7. By way of example, the Country Risk Experts assess that the country risk of Iraq (a UMIC in Group 7) requiring debt relief or defaulting on a loan is far higher than that of Botswana (another UMIC in Group 3)[16].
It would make far more sense for the DAC to use the Country Risk Categories published by the Export Credits Division than relying on World Bank Income categories.
Although the OECD does not publish the minimum premium rates that are calculated through this process, some National Export Credit Agencies do. The following implied risk spreads have been calculated based on the up-front risk-related premium rates that US EXIM Bank charges for an illustrative 10-year loan (see Annex 9) to a sovereign borrower, which conform with the Export Credit Arrangement’s disciplines. For the sake of comparison, this table shows how many countries from each World Bank income group fall in each country category.
Country risk category | Risk-related spread (MPRs) [17] | HICs (not ODA eligible) | UMICs: DAC 1% spread | LMICs: DAC 2% spread | LICs including LDCs: DAC 4% spread | |
1 | 0.25% | 1 | 0 | 0 | 0 | |
2 | 0.48% | 4 | 2 | 0 | 0 | |
3 | 0.79% | 3 | 7 | 4 | 0 | |
4 | 1.22% | 2 | 8 | 1 | 0 | |
5 | 1.73% | 1 | 9 | 6 | 0 | |
6 | 2.22% | 3 | 9 | 18 | 4 | |
7 | 2.85% | 1 | 13 | 15 | 24 |
Although produced for a different purpose (export credits rather that ODA loans), these risk premiums result from assessments of the same credit risks being proxied through the risk spreads implicit in the adjustment factors used in the DAC’s discount rates. Incidentally, this analysis drawing on premiums charged by US EXIM Bank consistent with the rules of export credits also suggests that the risk spreads used by the DAC are not inadequate to cover the credit risks of
ODA loans[18]. In fact, they seem rather excessive. What is certain is that because of the income category system used, the credit risk for some countries is being significantly over-priced, while for others the risk spread is inadequate.
Recommendation 3: Any adjustment to the discount rate to account for credit risk needs to reflect more accurately the different level of risks of borrowing developing countries.
Flaw 4: The reduction in the minimum grant element to qualify as ODA for loans to LMICs and UMICs means that low risk loans at quasi-commercial rates would now yield ODA credit
Under the previous system, the minimum grant element for a loan to qualify as ODA, regardless of the recipient country, was 25%. The new minimum requirement of 45% grant element for a loan to an LIC or LDC means that the donor must provide softer terms under the new methodology than previously for a loan to qualify as ODA[19].
However, for LMICs and UMICs, the minimum grant element has been reduced from 25% to 15% and 10% respectively.
Combined with the excessively high discount rates in the new methodology, this now makes it possible for a donor to provide loans at relatively high rates of interest, comparable with those of a commercial loan, and for the loan to yield ODA (see annex 7 for examples of how easy it is for even profitable loans to result in ODA credits for the donor).
The problem is not with the hurdle rate per se, but how it enables donors to take advantage of the high discount rates to register loans as ODA-eligible without “donor effort”.
Recommendation 4: The current rules stipulating minimum grant element for a loan to qualify as ODA should be re-examined in the context of broader reform to the methodology for counting ODA in concessional loans.
Flaw 5: Debt relief for non-ODA claims (OOF, export credits and private flows at market terms) is wrong: 1) many of these loans are not for development purposes and 2) debt relief involves no donor effort as it is funded by the developing countries themselves.
The DAC makes the following case for counting the costs of debt relief for non-ODA loans as new ODA:
“As no ODA grant equivalents will have been recorded for the original claims, the donor effort involved in concessional debt relief (forgiveness and concessional rescheduling) of non-ODA claims should give rise to ODA equivalents.”[20]
There are (at least) two major flaws in this reasoning.
The first is that many of these loans are for projects that do not meet the DAC’s own definition of ODA:
“Official development assistance (ODA) is defined by the OECD Development Assistance Committee (DAC) as having “the promotion of the economic development and welfare of developing countries as its main objective”[21].
Yet some of the loans may even be for “unproductive expenditure”[22] or other projects that do not contribute positively to development[23].
It is not clear why the costs associated with debt relief for loans that would not themselves qualify as development assistance (even if they met the grant equivalent criteria) should then count as ODA. As asked pithily by Jorge Rivera in his article, “Why the DAC’s latest move is bad for development”[24], “If loans were not for development purposes when they were first given, why should they be counted as aid when they are forgiven?”
Secondly, and perhaps even more fundamentally, recipient countries (the borrowers of these loans) pay risk premiums to cover the risk of debt relief and its costs, and (for export credit loans and guarantees at least) these have been set at rates that have been deemed by OECD governments as adequate to cover long term operating costs and losses (a requirement under the WTO Subsidies Code). Therefore, the DAC is effectively allowing donors to claim ODA credit for debt relief costs that have been funded through the risk premiums paid by commercial exporters, and invariably passed on to developing countries in the prices they pay for the goods and services supplied. Given there is no “donor effort” and no cost to the taxpayer, no ODA should be counted.
Recommendation 5: The practice of allowing the costs of debt relief for non-ODA loans where the estimated credit risk has been covered in the cost of such loans to the borrower should cease
V. Impacts of the Grant Equivalent System
The previous methodology of counting principal payments as ODA and then repayments of principal as negative ODA was fundamentally flawed, conceptually and in practice. And it would be wrong to say that there are NO benefits of the new system for counting ODA over the previous methodology:
The new system, unlike the old, does reward greater donor effort in one loan compared with another with more ODA credit (which wasn’t the case previously).
The higher minimum grant element requirement for a loan to an LIC/LDC to qualify as ODA does require greater effort from a donor to clear this hurdle than under the old methodology.
This paper is not intended as an attack on ODA loans. Indeed, it is recognised that they bring important benefits: they increase development country ownership; arguably improve the fiscal discipline of the recipient; can help to make scarce ODA funds go further; and enable the recycling of loan repayments into other ODA loans or grants[25]. And the previous methodology needed to be changed: the counting of ODA of a loan at its full face-value at disbursal but then deducting ODA for all repayments of principal meant that every ODA loan, regardless of the softness of its terms, always ended up yielding zero ODA. This unfairly disincentivised loans (at least over the long term).
However, the methodological flaws of the grant equivalent system that the DAC has introduced and that are outlined above have serious negative impacts.
Firstly, the donor effort of those countries that provide aid in the form of loans is being massively exaggerated, thereby fatally undermining any credibility in the ODA figures released by the OECD. Taxpayers are being seriously misled.
Secondly, many loans that currently qualify as ODA entail little or no donor effort (as would be shown if more rational discount rates were used). Indeed, the reduced minimum grant element requirement for loans to LMICs and UMICs raise the prospect of “donors” receiving ODA credit for extending loans on commercial terms.
Thirdly, the current system massively incentivises loans over grants: for the equivalent “donor effort” (in terms of cost to taxpayer), loans yield much more ODA than grants. Incentivised by the new system, bilateral loans have risen enormously since the Grant Equivalent methodology has been implemented. In its report about ODA in 2020 (a year in which total ODA rose by just 3.5% in real terms, the OECD noted that “Bilateral sovereign loans by DAC members on a grant equivalent basis increased by 38.7% in real terms compared to 2019.”[26]
Fourthly, at a time of rising global interest rates and increased debt stress among many developing and transition economies, arguably the DAC methodology (and loan-pushing by donors) is helping to fuel an impending debt crisis, and countries are being encouraged to over-borrow. According to the Jubilee Debt Campaign, 54 countries across the world are currently suffering from a debt crisis[27]. And the latest World Bank/IMF Debt Sustainability Analysis of Low Income Countries[28] shows that out of the 69 countries listed, 38 are either “in distress” or at “high risk” of overall debt distress, and a further 20 countries are at moderate risk of debt distress.
Fifthly, ODA is being counted with no justification for debt relief for export credits and other loans that may bring little or no developmental benefit when such debt relief has been funded by developing countries through the risk premiums they have paid. Accordingly, such debt relief entails no cost to the donor countries, no donor effort and is for projects that may have no development benefits.
Illustration of the Magnitude of Over-Counting
Without yet having details of ODA loans in 2020, it is not possibly to make an accurate assessment of the aggregate over-counting of ODA due to the flaws in the DAC methodology. However, the DAC reported[29] in April 2021 that the “Bilateral Grant equivalent of loans” in 2020 amounted to around US$ 13 billion. It seems safe to assume that the actual donor effort, measured using a discount rate that approximates to the cost of funds (and even adding a reasonable risk-related adjustment), is a small fraction of that figure.
Annex 7 illustrates the likely magnitude of the problem. Four loan scenarios are considered and the compared, all using the DAC’s own tool[30] for calculating ODA (only plugging in more robust discount rates):
i) 30-year loan from Japan to Indonesia for Yen 100 million (approximately Euros 780k) at an interest rate of 1% and 10-year grace period.
Under the DAC system, this yields ODA of 61.66% or more than Yen 61.5 million Yet the donor effort using the OECD’s DDRs is around just 17.54% or Yen 17.5 million Even if one adds the credit premium calculated by the OECD Country Risk Experts (which one shouldn’t in light of the DAC decision to count debt relief as additional ODA), the donor effort is just 30.32%[31].
ii) 10-year loan from Germany to India for Euros 5 million at an interest rate of 3% and a 1-year grace period.
DAC calculates this as yielding 17% ODA, more than Euros 850,000 Yet donor effort using DDR shows that this loan actually results in a net gain for the German taxpayer of 8.88%, pocketing some Euros 444k for the donor Adding the credit premium in this case still results in small positive grant element of 4.72%.
iii) 15-year loan from France to Algeria for Euros 1 million at an interest rate of 4% and a 1-year grace period
DAC calculates this as yielding 11.75% ODA, more that Euros 117,500 Using DDR donor is seen to get a net return (rather than cost) of 18.41%, bringing in around Euros 184k for the German taxpayer. Adding the risk premium, there is still “negative ODA” (i.e. A net gain to the taxpayer) of Euros 33,000
iv) 10-year loan from Spain to Honduras for Euros 4 million at an interest rate of 3% and a 1-year grace period (similar to the German loan above)
DAC calculates that this yields ODA of 17%, more than Euros 680k DDR shows that donor effort is again “negative” and Spanish taxpayers can expect a return of around Euros 355k. Due to the higher risk premium for Honduras, adding this to the discount rate yields a small cost to the donor of 1.45% or around Euros 58,000, but this is below the threshold to count as ODA.
Hence, using these examples, the DAC calculates that these loans, totalling just under Euros 11 million, yield aggregated ODA of around Euros 2.1 million. Yet, using a more robust discount rate (in this case the DDRs used for determining the concessionality level of tied aid), only one of the loans has a positive grant element (of just Euros 133k) that would count as ODA. This illustrates the magnitude of the over-counting.
While using some different lenders, recipients, interest rates and profiles would yield lower differences, some would show even bigger differences between the ODA yielded under the DAC system and true donor effort.
VI. Impact on Climate Finance
Perhaps of greatest concern is the impact of this methodology on climate finance.
In the run-up to COP26 in Glasgow, much attention was focused on the commitment of developed countries to mobilise US$ 100 billion a year by 2020 (originally agreed in the Copenhagen Accord back in 2009) to address the mitigation and adaptation needs of developing countries. There was strong criticism from different quarters[32] about the continuing failure of developing countries to hit their agreed target, as well as accusations that the figures published by the OECD included double-counting, “greenwashing” and showed the diversion of existing development resources away from critical areas for poverty reduction.
While the $100 billion is but a fraction of the resources needed by developing countries, it is a key element of the Paris Agreement, important both symbolically, as a clear commitment from developed countries to help finance the costs of climate change for countries less able to do so, and materially, as it can help to leverage the vast resources needed from various sources.
The Independent Expert Group on Climate Finance Report[33] was clear about the importance of bilateral finance in the context of this commitment:
“Bilateral climate finance is the source of almost all commercial climate finance and is vital to shore up the integrity, solidity and predictability of finance flows for adaptation and mitigation action to support developing countries, and to help scale up other sources of finance.”
However, like other commentators, the Expert Group regretted the dwindling proportion of climate finance provided in the form of grants, even to the poorest countries, compared with loans. It stated that the figure of just over $12 billion in grants “falls far short of what is needed”; and argued that this figure should ideally treble by 2025. And, on the eve of COP26, the UK Parliament issued a note[34] lamenting that most public finance for climate change is now provided as loans, recording that “loans have increased from US$19.8 billion in 2013 to US$44.5 billion in 2019, 71% of the public climate finance provided in 2019”.
In its Climate Finance Shadow Report 202035, Oxfam also criticised the growing proportion of climate finance provided in the form of loans rather than grants, noting the ethical and financial issues:
“The world’s poorest countries and communities should not be forced to take out loans to protect themselves from the excess carbon emissions of rich countries. Finance that should be helping countries respond to climate change should not be harming them by contributing to rising – and in many cases, unsustainable – debt levels”.
It does indeed seem wrong to saddle developing countries with debts for costs of adapting to a climate crisis not of their making and for cleaning up the damage it causes. They may well ask where is the “polluter pays principle” here?
Of course, the DAC is not responsible for the broader concerns around climate finance, including the lack of a clear definition in the UNFCCC of what counts towards the $100 billion. However, the DAC is responsible for setting up a system where donors are financially incentivised to extend loans rather than grants, by allowing them to claim significant ODA for little or no effort from their taxpayers.
Indeed, while Oxfam has “estimated 40% of public climate finance is non-concessional”, that figure would be far higher if a lower discount rate, reflecting more accurately true donor effort, were used to calculate concessionality.
Moreover, the preponderance of ODA loans (instead of grants) being extended as climate finance makes blended finance and the involvement of the private sector more difficult, as developing countries are becoming “over-extended” and suffering from debt stress through the ODA loans they are having to service.
While it cannot be definitively shown that the growth of Governmental loans rather than grants in climate finance is a direct result of the DAC’s grant equivalence system, the new methodology certainly provides an incentive for governments to provide ODA in this way.
The result is that developing countries are being made to pay for a climate change that they did not cause, while OECD countries are claiming credit for aid that they have not given. This seems to be the antithesis of climate justice.
VII. A Different Approach
In light of the foregoing, firstly a new method for calculating the grant equivalent in loans needs to be instituted urgently, based on a discount rate that accurately reflects donor effort.
In the early 2000s, the OECD development community spoke a great deal about the need for more “policy coherence”. Indeed, I led the creation of the Organisation’s horizontal work programme on Policy Coherence for Development. In this connection, the DDRs of the OECD, used for the purpose of calculating the concessionality of tied aid, are already accepted by OECD governments as a good proxy for the funding costs of government (in fact, with the addition in the calculation of the CIRRs of 100 basis points to long term bond yields[36], even these are arguably too high). Adoption of discount rates based on the DDRs would reduce the “incoherence” of the current OECD approach to counting ODA in loans.
With respect to credit risk, a decision is needed either to price this in the discount rate OR to allow the costs of debt relief to count as ODA. You cannot do both. If the current practice of allowing debt relief costs to count as ODA is changed, a new methodology for pricing credit risk in the discount rate needs to draw on the work undertaken elsewhere in the OECD and by professional credit risk agencies to ensure the margin reflects more accurately the risks.
Finally, the policy of counting the costs of debt relief related to non-developmental loans should be stopped.
VIII. The Road to Change and the Role of the DAC
The impact on climate finance, at this critical juncture, means that this issue cannot be left to the DAC review scheduled for 2023. It needs to be addressed now. And the international community needs to reflect carefully and urgently whether a body comprising only donor countries has revealed an institutional conflict of interest in the way that it has exaggerated so massively ODA in loans.
Taken together, the shortcomings of the grant equivalent methodology adopted by the DAC, combined with the decision to add ODA for the costs of debt relief for both ODA and non-ODA result in the systematic inflation of ODA statistics, making donors appear far more generous than they are. It is hard to believe that the most egregious flaws in the ODA methodology are not known to the DAC Secretariat, especially as other OECD colleagues have developed robust mechanisms for calculating the concessionality level of tied aid (since 1992) and for calculating country credit risk for medium- and long-term export credits (since 1997).
Of course, it is recognised that the determination of what constitutes ODA and the methodology for how it should be counted are consensus decisions made by the members of the DAC. But this surely raises the most fundamental question: why is the definition and counting of ODA tasked to a committee comprising only donor countries, who are, arguably, collectively “incentivised” to demonstrate their largesse and generosity in terms of the aid they provide, without any developing countries in attendance to ensure accuracy and robustness? In an age of supposed country ownership for their own economic development, this looks anachronistic. Even if one doesn’t consider the DAC to have an institutional conflict of interest in calculating ODA given its membership, it is hard not to conclude that it has forfeited its authority in this area given its inability to put in place a robust methodology for doing so.
Recommendation 6: The task of defining and calculating ODA should be considered by an experts’ group with the requisite knowledge and experience to analyse the relevant issues and develop a sound methodology drawing on international best practice. Implementation and reporting could then be passed to an independent organisation that is free from political control.
Note about the Author:
Stephen Cutts worked as an insurance underwriter and reinsurance broker in the City of London before becoming a Fast Stream Civil Servant and joining the UK’s Export Credits Guarantee Department (now UK Export Finance) in 1994. As Head of International Relations, Development, Steve led the UK Government’s negotiations at the OECD and in the EU on risk-related pricing and other export credits issues, seeking to ensure a level playing field for exporters.
Recruited to the OECD in 1996, Steve led negotiations for the Secretariat on various export credits agreements, securing the first agreements among governments to take account of the environmental impact of the projects they were supporting, and to take measures to help eliminate bribery from projects benefiting from government support.
Joining the Secretary-General’s Office in 2000, Steve led the development of (and raised the funding to implement) the OECD’s programme on Policy Coherence for Development to help ensure that development issues were duly considered across the somewhat siloed policy committees of the Organisation.
Promoted to Chef-de-Cabinet/Chief of Staff in 2004, Steve became the chief adviser to the Secretary-General of the time, Donald Johnston, until SG Johnson retired from the OECD in 2006.
Following a brief stint as Executive Director of the Africa Programme at INSEAD, Steve joined the Commonwealth Secretariat in 2008, serving first as Director of Strategic Planning and Evaluation and then becoming the inaugural Assistant Secretary-General for Corporate Affairs. At the Commonwealth, Steve led a major reform programme, including developing a new scale of assessed contributions thereby placing the organisation onto a sustainable financial footing, as well as modernising its HR, financial and operational policies and practices. In 2012, he led the development and negotiations of the Commonwealth Charter, which was signed by Her Majesty the Queen.
As Assistant Secretary-General for Central Support Services at the United Nations Secretariat (2013-2017), Steve instigated reforms in Procurement, Travel and Property Management to streamline the cumbersome bureaucracy of the United Nations and make the organisation more fit for purpose. He also led the design and implementation of flexible workplace at the UN Secretariat, saving the organisation more than US$20 million per annum by reducing the need for commercial rental space; ensuring that office space at UN HQ was configured to meet the working needs of staff; and providing tools to enable personnel to work remotely.
Steve left the UN in 2017 to lead an international education charity promoting improvements in foundational literacy and numeracy in some of the poorest remote villages in India. Today, he shares his time between Sussex in the UK and New York, serving as a trustee on a number of not-for-profit boards and pursuing a range of interests.
Annex 1 The Previous Method for Calculating the ODA of Concessional Loans
Until 2018, the DAC calculated the ODA in concessional loans using a cash flow method, counting outflows and inflows of principal. Under this method, the full-face value of a loan was counted as ODA on disbursement and repayments of principal (ignoring payments of interest) were then counted as negative ODA.
The interest rate of such loans was only taken into account in the calculation of the “grant element”, which was calculated using a discount rate of 10% (i.e. the Net Present Value of all future repayments of principal and interest were discounted at 10%), and the grant element was deemed to be the level by which the NPV of all cashflows fell short of the full face value of the loan. The “hurdle rate” (the minimum percentage of the grant element represented against the full-face value of the loan to qualify as ODA) was set at 25%. Having cleared that hurdle, both interest rate payments and the grant element were in the calculation of ODA.
This methodology was fundamentally flawed for host of reasons, including:
It ignored the “time value of money”, taking no account of the fact that, over time, money gradually loses its value.
It did not distinguish any difference in “donor effort” between highly concessional loans (say a 20-year loan with zero interest rate and a bullet repayment at year 20) and loans on much less generous terms (for example, a loan over the same 20 years but with an interest rate of 7% requiring equal annual repayments of principal).
It inflated the ODA at the time the loan was extended (by counting the full face-value, it yielded the same ODA as a grant) yet when fully repaid, every concessional loan, regardless of its terms, counted as zero ODA (as each repayment of principal counted as negative ODA until the loan was fully repaid).
The 10% discount rate was far too high to represent either the cost of funds (to donor governments) or even a commercial borrowing rate for the governments of most developing countries.
The common discount rate of 10% also took no account of significant differences in government financing and lending rates across different currencies, nor the significant difference in credit risk among developing country recipients.
Annex 2: Treatment of Debt Relief (Commentary on DAC note on “Reporting on Debt Relief in the Grant Equivalent System[37]):
I. Treatment of Debt Relief for ODA Loans
The DAC paper sets out the decision of the DAC in 2020 that additional concessionality on an ODA loan is counted as additional ODA, even though ODA has already been credited for the credit risk (through the additional 1%, 2% or 4% added to the 5% basic discount rate, depending the World Bank income category within which a recipient of an ODA loan falls).
The only caveat/limitation to this decision is that a ceiling is applied so that the new ODA for the additional costs resulting from the debt relief when added to the ODA originally recorded for the loan cannot exceed the original face value of the loan. In other words, the total ODA for a loan can never exceed the ODA value of a grant of an amount equivalent to the full face-value of the loan, regardless of the level or timing of any debt relief.
This caveat does not negate the fact that this is clear double-counting (even though the preamble in the Narrative in section 1.2 in the note recalls that in 2014, the DAC had “had agreed that the cost of risk should…not be double counted”). If the DAC wants to count donor effort in debt relief at the time such costs are borne, there is no justification for adding a credit risk spread to the discount rate.
The justification in the Narrative in section 1.2 of the DAC seeks to convince the reader that the practice of double-counting has some validity. It does not.
The first bullet states that the policy is aligned with the call of the 2014 HLM to bear in mind previous needs to encourage debt relief initiatives. The implication seems to be that counting the costs of debt relief as additional ODA offers an incentive to donors to grant such relief. The objective of incentivizing debt relief at the time this is needed by offering to credit ODA would be met equally well if no credit margin were added to the discount rate and the costs (as is the case now) were simply added to ODA as and when they occur. This alternative approach would avoid the double-counting as the 2014 HLM had urged.
The second bullet argues that loans and debt relief are legitimate means of delivering ODA. That is not questioned here. Indeed, the paper “Give Credit Where Credit’s Due” makes clear the benefits than ODA loans can bring. But there are also advantages in providing grants. The double-counting of allowing additional ODA for debt relief while simultaneously adding ODA (via an increased discount rate) up-front for the credit risk exacerbates the exaggeration of donor effort in providing loans over grants (which is already overstated due to the 5% base factor in the discount rate that far exceeds the cost of funds for donors).
The third bullet explains the limitation that it is not possible for any single loan to yield more ODA than a grant equal to the face value of the loan. However, this limitation neither negates nor significantly offsets the double-counting[38].
The fourth bullet claims that “the differential discount rates do not cover for the “cost of default” ex-post, which entails additional donor effort”. The argument here is disingenuous: the loan was extended in the full knowledge that there WAS a risk of default (otherwise a risk adjustment would not have been added to the discount rate). Of course, it is true that defaulted loan entails more effort than a loan that does not default, but the default risk has already been accounted for in the risk-related adjustment to the discount rate. This argument in the paper ignores the fundamental principle of insurance around spread of risk and that premiums across a portfolio of risks are to pay for losses. Again, the need for debt relief for a particular loan does not mean that the ex-ante risk spread was wrong.
This is like telling a householder whose property has burned to the ground that the insurance premiums s/he has paid are less than the cost of the damage sustained and that, therefore, the policy holder will have to pay the claim…minus the value of the premiums s/he has paid (so that the total amount the householder pays won’t be more than the cost he would have borne if the property had burned down on day one, and no insurance had been in place).
The bullet continues, “The occurrence of debt relief means that the context has changed in comparison with the original decision to extend the loan.”
A risk adjustment was made to the discount rate precisely because the lender knew there was a chance that debt relief would be needed. What about all the loans that don’t require debt relief? Arguably, their context has also changed as the credit risk added to the discount rate proved unnecessary. If the DAC wants to add the cost of debt relief to ODA as well as making a credit risk adjustment to the discount rate, it should subtract, ex post facto, the risk-related portion of ODA included in the discount rate from every loan that does not incur any costs associated with debt relief.
Finally, this fourth bullet implies that ODA loans, due to their long repayment schedules, might carry additional risks:
“ODA loans may have long maturities over which it is not possible to foresee all possibilities – conflicts or natural disasters may occur; other external factors such as volatility in resource prices or contingent liabilities unknown to the creditor may also lead to unforeseen difficulties for the borrower to repay its loan.”
Of course, the longer the term of a loan (like any insurance policy), the more likely a default (or claim) will occur. However, this does not mean that the risk is increasing over time. The wording in the DAC paper suggests that the risk of default will be higher in, say 20 years, that it is in, say, five years. Although certainly less foreseeable, there is no reason to assume escalating risk. Indeed, one could argue that the opposite should be assumed for a majority of developing economies. After all, the purpose of development assistance is, at least in part, to both grow and increase the resilience of economies in developing and transition economies, so that they can better withstand shocks and avoid the need for debt relief. Accordingly, surely the development community should assume that their contribution is reducing the year-on-year risk of default of most developing countries so that it should be lower, say, in 2050, than it will be in 2030.
The fifth bullet states that the “risk-adjusted discount rates will be regularly reviewed”. A review of the rates is beside the point here - although of course, risk premium rates should be reviewed regularly in light of claims experience. In fact, a comparison with the premium rates for export credits that DAC Governments have determined are adequate to cover costs and losses suggest the risk spreads in the ODA discount rates may be too high. In any event, uncertainty about whether current discount rate spreads are commensurate with the risks they are intended to cover does not justify the current practice. Counting debt relief when it happens as ODA obviates the need for ANY risk spread in the discount rate.
The sixth bullet attempts again to justify why costs associated with debt relief of a particular loan should be added to ODA: “The method for accounting debt relief in a grant equivalent system aims at quantifying [the] additional concessionality by calculating the new grant equivalent of the loan, post treatment, and deducting the original grant element.” Of course, a loan requiring debt relief incurs additional costs for the donor, but to calculate this and add it to ODA is (again) asymmetric treatment, as there is no reduction of the ODA for all the loans that do not require debt relief. The adequacy or otherwise of the credit risk spread needs to be considered over time and over the full portfolio of loans. But again, if the costs of debt relief are simply to be counted as additional ODA, there is (again) no justification for adding a credit risk margin to the discount rate up-front.
The seventh bullet acknowledges the risks of freezing the credit risk margins of discount rates based on WB country group classifications, noting that the market would make periodic adjustments to account for changing risks. As noted elsewhere in the paper, the DAC seems to have overlooked the work undertaken by OECD Member Countries in the context of the Export Credit Arrangement in pricing non-repayment risks and placing developing and emerging economies from across the globe into eight risk categories[39]. This work – as well as a review of the findings of any of the major credit risk agencies (Standard and Poors, Moodys, Fitch etc.) - shows that the DAC’s approach of assigning the same credit risk to all countries in the same World Bank income category is deeply flawed (see separate section of the paper).
This bullet also notes that the Secretariat plans a first review of the ODA discount rates in 2023. This review should not wait and should be undertaken by an independent body given the serious flaws in the current methodology and the impacts they are having.
Finally, the eighth bullet commits the Secretariat to “monitor closely the implementation of the grant equivalent system…[and] the impact of the debt relief methodology.” The DAC paper notes that under its methodology there is a risk of members “circumventing the rules (instead of debt relief, providers give grants for borrowing countries to repay their loans).” It should be noted that should the alternative explored in this paper be adopted – that debt relief costs be counted as ODA instead of (rather than as well as) including an additional risk spread in the initial discount rate - this risk would be effectively mitigated for ODA loans as either approach (counting the costs of the debt relief as ODA or giving a grant of equal value) would yield identical additional ODA.
II. Treatment of Debt Relief for Non-ODA claims (OOF, export credits and private flows at market terms)
The DAC makes a simple case for counting the costs of debt relief for non-ODA loans as new ODA:
“As no ODA grant equivalents will have been recorded for the original claims, the donor effort involved in concessional debt relief (forgiveness and concessional rescheduling) of non-ODA claims should give rise to ODA equivalents.”
There are (at least) two major flaws in this reasoning.
The first is that many of these loans are for projects that do not meet the DAC’s own definition of ODA:
“Official development assistance (ODA) is administered with the promotion of the economic development and welfare of developing countries as its main objective.[40]”
Some of the loans may even be for “unproductive expenditure” or military goods.
It is not clear why the costs associated with debt relief should qualify as ODA when the underlying loans would not have qualified as an ODA loan, even if they had met the grant equivalency criteria.
Secondly, and perhaps even more fundamentally, recipient countries (the borrowers of these loans) pay risk premiums to cover the risk of debt relief and its costs, and (for export credit loans and guarantees at least) these have been set at rates that are adequate to cover long term operating costs and losses as this is a requirement on all export credit agencies under the WTO Subsidies Code. Therefore, the DAC is effectively allowing donors to claim ODA credit for debt relief costs that are paid from the funds collected from developing countries, with no “donor effort” whatsoever.
The practice of allowing donors to count ODA for debt relief of these instruments is justifiable neither on development nor on donor effort grounds.
Annex 3: Three Options for ODA Accounting for Credit Risk/Debt Relief.
Option 1:
Include a risk-related adjustment to the discount rate up-front (current practice), with no additional ODA for actual debt relief
Arguments For:
No double counting as the costs of the debt relief should have been covered in the up-front fee.
Provides for more market-comparable discount rates (a commercial lender would always price in the risk up-front).
Challenges
This does not provide an additional incentive on donor countries to provide debt relief (although it should be noted that they have already received ODA credit for this risk up-front)
Experience would need to be monitored over time to ensure the inferred “credit risk premium” included in the discount rate was neither excessive nor inadequate to cover long-term losses (it is understood that such monitoring is undertaken by the Export Credit Participants with respect to the Minimum Premium Rates cited in this paper).
There is an “ODA allocation” issue in that those countries that benefit from debt relief are not shown as having received any more ODA than those that haven’t. This is a reporting challenge - a country that has its debts forgiven has received more ODA than those countries that repay their loans in full. However, this does NOT justify donors counting the forgiveness as additional ODA that they have given - the “loss” has already been accounted as ODA through the risk spread across all the loans extended. There is no additional donor effort, provided the margin in the discount rate is commensurate to the credit risk.
Option 2:
No up-front adjustment to the discount rate for credit risk (discount rate to be based on donor cost of funds only). Costs of debt relief added when defaults/rescheduling/forgiveness occurs
Arguments For:
Accurate accounting of donor effort as ODA and no need to review assumptions about credit risk in light of loss experience. (After all, the actual and full costs of debt relief are recorded as ODA at the time the costs are incurred, so there is no additional donor effort that would warrant ODA being counted additionally through an adjusted up-front discount rate.)
The ODA figures accurately reflect the benefit that each developing country receives through debt rescheduling/forgiveness. In other words, the “ODA allocation” issue under Option 1 is avoided.
This is consistent with the treatment of tied aid, as administered by the OECD. The Differentiated Discount Rates for calculating the concessionality level of tied aid include no margin for the credit risk of non-repayment by the borrower.
Challenges
Intuitively, it can look strange to apply the same discount rate to loans for countries with very different credit risks (in fact, this is because the ODA budget is, separately, acting like an unconditional guarantor of the loan – it will fund any losses, so there’s no need to price in credit risk)
There is an administrative challenge with this option. In order to avoid under-counting ODA, it would be important to monitor loans where the interest rates neither include the full credit risk in their interest rates, yet nor do they qualify as ODA loans (as their grant element does not meet the ODA eligibility criteria)[41]. In the event of debt relief, there is donor effort for these loans that has not been fully offset by a risk-related premium/spread. This might require the maintenance of a simple shadow database of such loans, detailing their grant equivalent, so that appropriate ODA can be counted should debt relief be needed..
Option 3 (Based on Current Practice):
Current Practice: Include an adjustment for credit risk in the discount rate, and then add additional ODA to reflect the donor effort in affording debt relief for individual loans
Arguments For:
Ensures no under-counting of ODA due the credit risk-related adjustment to the discount rate being inadequate to cover real losses.
Ensures that ODA is “allocated” to those countries which benefit from debt rescheduling or forgiveness.
Challenges
This involves significant double-counting of ODA, giving credit for the assumption of the credit risk up-front, and again when losses are realized.
In order to avoid the current asymmetric treatment of only adjusting the ODA for loans that require debt relief, it would be necessary to deduct ODA in respect of all the loans that do not require debt relief – after all, the adjustment to the discount rate for credit risk proved to be unnecessary. This would be complex, cumbersome and, intuitively, would end up in the same situation as option 2 (only through much more administrative effort).
Overall Conclusion:
There is no option that does not entail an administrative challenge.
The choice between Option 1 and Option 2 depends on the weight one places on the need to incentivise debt relief compared with the desire for simplicity.
Option 3 is the most complex and administratively burdensome to manage if done properly, i.e. without the current asymmetric treatment, yet offers little (if any) benefit over Option 2.
There is need for further work whichever option is selected, to ensure clarity and transparency of reporting without sacrificing the integrity of ODA figures based on true donor effort.
Annex 4 Credit Risk: Country Classifications and Appropriate Risk Spreads in the Discount Rate
Measuring credit risk is never an exact science, but governments that provide official export credit support are required under the subsidies code of the WTO to ensure that they collect risk-related premiums that are sufficient to cover the risk of future losses.
Item j of the Illustrative List of Export Subsidies prohibited under the WTO Agreement on Subsidies and Countervailing Measures:
“The provision by governments (or special institutions controlled by governments) of export credit guarantee or insurance programmes, of insurance or guarantee programmes against increases in the cost of exported products or of exchange risk programmes, at premium rates which are inadequate to cover the long-term operating costs and losses of the programmes.”
Given this requirement, Country Risk Experts (CREs) representing the Participants to the Arrangement on Officially Support Export Credits developed a Model some 25 years ago for assessing the five elements of country risk for virtually all developing countries (text from the Arrangement paraphrased below):
general moratorium on repayments decreed by the obligor’s/guarantor's government or by that agency of a country through which repayment is effected;
political events and/or economic difficulties arising outside the country of the notifying Participant or legislative/administrative measures taken outside the country of the notifying Participant which prevent or delay the transfer of funds paid in respect of the credit;
legal provisions adopted in the obligor’s/guarantor’s country declaring repayments made in local currency to be a valid discharge of the debt, notwithstanding that, as a result of fluctuations in exchange rates, such repayments, when converted into the currency of the credit, no longer cover the amount of the debt at the date of the transfer of funds;
any other measure or decision of the government of a foreign country which prevents repayment under a credit; and
cases of force majeure occurring outside the country of the notifying Participant, i.e. war (including civil war), expropriation, revolution, riot, civil disturbances, cyclones, floods, earthquakes, eruptions, tidal waves and nuclear accidents.
The CREs classify countries into one of eight Country Risk Categories (0-7). Risk-related minimum premium rates have been established for Categories 1 through 7, but not for Category 0, as the level of country risk is considered to be negligible for countries in this Category (which are predominantly the High-Income OECD countries).
c) The classification of countriesis achieved through a methodology comprising:
A Country Risk Assessment Model (the Model), which produces a quantitative assessment of country credit risk which is based, for each country, on three groups of risk indicators: the payment experience of the Participants, the financial situation and the economic situation.
The methodology of the Model consists of different steps including the assessment of the three groups of risk indicators, and the combination and flexible weighting of the risk indicator groups.
The qualitative assessment of the Model results, considered country-by-country to integrate the political risk and/or other risk factors not taken into account in full or in part by the Model. When deemed appropriate, these lead to an adjustment to the quantitative Model assessment to reflect the final assessment of the country credit risk.
Country Risk Classifications are monitored on an on-going basis and reviewed at least annually and changes resulting from the Country Risk Classification Methodology are immediately communicated by the Secretariat.
The country risk classifications are made public by the OECD Secretariat; the latest listing is here: https://www.oecd.org/trade/topics/export-credits/documents/cre-crc-current-english.pdf
The minimum premium rates are, however, NOT made public by the Secretariat. Nevertheless, a number of Export Credit Agencies do provide public calculators that effectively give these rates. Here are links to US EXIM Bank’s calculator and explanatory note
The premium fees are expressed on an upfront basis. For a sovereign borrower of a 10-year loan requiring with immediate drawdown, the US EXIM Bank calculator reveals that the Minimum Premium Rates (unfinanced) for each of the seven risk categories are as follows:
Country risk category | Upfront premium |
1 | 1.32% |
2 | 2.50% |
3 | 4.13% |
4 | 6.37% |
5 | 9.05% |
6 | 11.60% |
7 | 14.92% |
To equate these to a risk-based spread (so that the DAC risk spreads of 1%, 2% and 4% can be compared), we applied current OECD rates[42], using “Convention B” to determine the discount rate applicable to premium fees collected after the starting point of credit for the US Dollar[43] (see also Annex 9).
This shows that the equivalent risk spreads would be:
Country risk category | Equivalent risk spread |
1 | 0.25% |
2 | 0.48% |
3 | 0.79% |
4 | 1.22% |
5 | 1.73% |
6 | 2.22% |
7 | 2.85% |
This suggests that the DAC is overstating the credit risk in its spreads added to the base factor in the discount rates it applies for the calculation of ODA.
Moreover, a closer look at the countries in the different credit risk categories suggests another major problem with the DAC methodology.
The DAC approach of equating the different income categories of countries with their credit risk is flawed. In fact, the data from the country risk experts involved in the risk pricing of export credit loans does not show a strong correlation. While it is true that all LICs and LLDCs fall within Categories 6 and 7, suggesting a discount rate of around 3% maximum (not 4%) is justifiable, UMICs and LMICs are scattered across the different risk categories. Specifically, UMICs are found in Group 2 through 7 and LMICs are found in Groups 3 through 7. This demonstrates that it is not appropriate to assign a single discount rate to all countries in a specific World Bank income category.
By way of example, the Country Risk Experts assess that the country risk of Iraq (a UMIC in Group 7) is far higher than that of Botswana (Group 3) and warrants nearly four times the premium rate.
The ODA calculation needs to take a far more differentiated approach, based on assessed credit risk, and apply different discount rates for loans to different developing countries to reflect, more accurately the risk of default, debt rescheduling and debt forgiveness. Otherwise, some of the estimations of donor effort and grant equivalence will be highly inaccurate. Moreover, the current overestimation of risk in the adjustments needs to be corrected.
Recognising the desire for a simple system for taking account of credit risk in the discount rate, this paper applies the following adjustments for risk based on the findings above:
Country risk category | Proposed adjustment |
1 | Not needed as no ODA eligible countries in this Category |
2 | 1% |
3 | 1% |
4 | 1% |
5 | 2% |
6 | 2% |
7 | 3% |
Annex 5 Prevailing Country Risk Classifications (from the Country Risk Methodology of the Export Credit Arrangement)
Group 1
Country | Income classification (2022) |
Chinese Taipei | HIC |
Group 2
Country | Income classification (2022) |
China (People's Republic of) | UMIC |
Hong Kong (China) | HIC |
Kuwait | HIC |
Malaysia | UMIC |
Saudi Arabia | HIC |
United Arab Emirates | HIC |
Group 3
Country | Income classification (2022) |
Azerbaijan | UMIC |
Botswana | UMIC |
Bulgaria | UMIC |
India | LMIC |
Indonesia | LMIC |
Mauritius | UMIC |
Mexico | UMIC |
Morocco | LMIC |
Peru | UMIC |
Philippines | LMIC |
Qatar | HIC |
Romania | UMIC |
Thailand | UMIC |
Trinidad and Tobago | HIC |
Uruguay | HIC |
Group 4
Country | Income classification (2022) |
Bahamas | HIC |
Colombia | UMIC |
Costa Rica | UMIC |
Croatia | HIC |
Dominican Republic | UMIC |
Guatemala | UMIC |
Panama | UMIC |
Russia | UMIC |
Serbia | UMIC |
South Africa | UMIC |
Viet Nam | LMIC |
Group 5
Country | Income classification (2022) |
Albania | UMIC |
Algeria | LMIC |
Bangladesh | LMIC |
Brazil | UMIC |
Cote d'Ivoire | LMIC |
Curacao | HIC |
Egypt | LMIC |
Fiji | UMIC |
Honduras | LMIC |
Jordan | UMIC |
Kazakhstan | UMIC |
Northern Macedonia | UMIC |
Paraguay | UMIC |
Senegal | LMIC |
Turkey | UMIC |
Uzbekistan | LMIC |
Group 6
Country | Income classification (2022) |
Angola | LMIC |
Armenia | UMIC |
Aruba | HIC |
Bahrain | HIC |
Belarus | UMIC |
Benin | LMIC |
Bhutan | LMIC |
Bolivia | LMIC |
Cabo Verde | LMIC |
Cambodia | LMIC |
Cameroon | LMIC |
Ecuador | UMIC |
El Salvador | LMIC |
Eswatini | LMIC |
Gabon | UMIC |
Georgia | UMIC |
Ghana | LMIC |
Guyana | UMIC |
Jamaica | UMIC |
Kenya | LMIC |
Kosovo | UMIC |
Lesotho | LMIC |
Namibia | UMIC |
Nepal | LMIC |
Nigeria | LMIC |
Oman | HIC |
Papua New Guinea | LMIC |
Rwanda | LIC |
Tanzania | LMIC |
Timor-Leste | LMIC |
Togo | LIC |
Tunisia | LMIC |
Uganda | LIC |
Ukraine | LIC |
Group 7
Country | Income classification (2022) |
Afghanistan | LIC |
Antigua and Barbuda | HIC |
Argentina | UMIC |
Bosnia and Herzogovina | UMIC |
Burkina Faso | LIC |
Burundi | LIC |
Central African Republic | LIC |
Chad | LIC |
Congo | LMIC |
Cuba | UMIC |
Democratic People's Republic of Korea | LIC |
Democratic Republic of the Congo | LIC |
Djibouti | LMIC |
Equatorial Guinea | UMIC |
Eritrea | LIC |
Ethiopia | LIC |
Gambia | LIC |
Guinea | LIC |
Guinea-Bissau | LIC |
Haiti | LMIC |
Iran | LMIC |
Iraq | UMIC |
Kyrgyzstan | LMIC |
Lao People's Democratic Republic | LMIC |
Lebanon | UMIC |
Liberia | LIC |
Libya | UMIC |
Madagascar | LIC |
Malawi | LIC |
Maldives | UMIC |
Mali | LIC |
Mauritania | LMIC |
Moldova | UMIC |
Mongolia | LMIC |
Montenegro | UMIC |
Mozambique | LIC |
Myanmar | LMIC |
Nicaragua | LMIC |
Niger | LIC |
Pakistan | LMIC |
Sierre Leone | LIC |
Somalia | LIC |
South Sudan | LIC |
Sri Lanka | LMIC |
Sudan | LIC |
Suriname | UMIC |
Syrian Arab Republic | LIC |
Tajikistan | LMIC |
Turkmenistan | UMIC |
Venezuela | UC/UMIC |
West Bank and Gaza | LMIC |
Yemen | LIC |
Zambia | LMIC |
Currently not reviewed or clasified
Country | Income classification (2022) |
Barbados | HIC |
Belize | LMIC |
Brunei Darussalem | HIC |
Comoros | LMIC |
Dominica | UMIC |
Kiribati | LMIC |
Liechtenstein | HIC |
Macau (China) | HIC |
Marshall Islands | UMIC |
Micronesia | LMIC |
Nauru | HIC |
Palau | HIC |
Saint Kitts and Nevis | HIC |
Saint Lucia | UMIC |
Saint Vincent and the Grenadines | UMIC |
Samoa | LMIC |
Sao Tome and Principe | LMIC |
Seychelles | HIC |
Sint Maarten | HIC |
Solomon Islands | LMIC |
Tonga | UMIC |
Tuvalu | UMIC |
Vanuatu | LMIC |
Annex 6 Comparison of DAC’s Income-Based Approach and the Export Credit Arrangement’s Risk Categories and Risk Spreads
Country risk category | Risk-related spread (MPRs) | HICs (not ODA eligible) | UMICs: DAC 1% spread | LMICs: DAC 2% spread | LICs including LDCs: DAC 4% spread | Proposed simplified spread |
1 | 0.25% | 1 | 0 | 0 | 0 | N/A |
2 | 0.48% | 4 | 2 | 0 | 0 | 1% |
3 | 0.79% | 3 | 7 | 4 | 0 | 1% |
4 | 1.22% | 2 | 8 | 1 | 0 | 1% |
5 | 1.73% | 1 | 9 | 6 | 0 | 2% |
6 | 2.22% | 3 | 9 | 18 | 4 | 2% |
7 | 2.85% | 1 | 13 | 15 | 24 | 3% |
The Risk-Related spreads were calculated from the up-front premium fees from the US EXIM Bank calculator (see Annex 4) and are consistent with the Minimum Premium Rates of the Participants to the Export Credits Arrangement.
Annex 7: A Comparison of DAC Calculations with Donor Effort (all calculations using DAC Tool[44]).
All assume equal annual repayments of principal
Based on the risk spread equivalents (see annex 4) but also the needs for simplicity, these
calculations apply risk spreads of 1% for Country Risk Categories 2, 3 and 4; 2% for Country Risk
Categories 5 and 6; and 3% for Category 7
Loan 1:
Donor: Japan
Recipient: Indonesia
Loan Value: Yen 100 million
Repayment term: 30 years
Grace Period: 10 years
Interest Rate: 1%
DAC Discount Rate: 7% (LMIC)
DDR: 2.1%
Minimum Risk Spread: 1% (Country Risk Category 3)
DAC Calculation of ODA: 61.6579% (Yen 61.6579 million)
Donor Effort using DDR: 17.5398 % (Yen 17.5398 million)
[Adding Risk Spread[45]: 30.3240% (Yen 30.324 million)]
Loan 2:
Donor: Germany
Recipient: India
Loan Value: Euros 5 million
Repayment term: 10 years
Grace period: 1 year
Interest Rate: 3%
DAC Discount Rate: 7% (LMIC)
DDR: 3%
Minimum Risk Spread: 1% (Country Risk Category 3)
DAC Calculation of ODA: 17.0081% (Euros 850,405)
Donor Effort using DDR: -8.8836% (minus Euros 444,180)
[Adding Risk Spread: 4.7228% (Euros 236,140)]
Loan 3:
Donor: France
Recipient: Albania
Loan Value: Euros 1 million
Repayment term: 15 years
Grace Period: 1 year
Interest Rate: 4%
DAC Discount Rate: 6% (UMIC)
DDR: 1.5%
Minimum Risk Spread: 2% (Country Risk Category 5)
DAC Calculation of ODA: 11.7506% (Euros 117,506)
Donor Effort using DDR: -18.4055% (minus Euros 184,085)
[Adding Risk Spread: (-3.3168% (minus Euros 33,168)]
Loan 4:
Donor: Spain
Recipient: Honduras
Loan Value: Euros 4 million
Repayment term: 10 years
Grace period: 1 year
Interest Rate: 3%
DAC Discount Rate: 7% (LMIC)
DDR: 1.3%
Minimum Risk Spread: 2% (Country Risk Category 5)
DAC Calculation of ODA: 1 17.0081% (Euros 680,324)
Donor Effort using DDR: -8.8836% (minus Euros 355,344)
[Adding Risk Spread: 1.4536% (Euros 58,144)]
Annex 8 Differentiated Discount Rates (DDR) Effective as from 15 January 2022
Currency | Code | R < 15 years | 15 ≤ R < 20 years | 20 ≤ R < 30 years | R ≥ 30 years |
Margin (included) | 0.75 | 1.00 | 1.15 | 1.25 | |
Australian Dollar | AUD | 2.8 | 3.0 | 3.2 | 3.3 |
Canadian Dollar | CAD | 3.0 | 3.2 | 3.4 | 3.5 |
Czech Koruna | CZK | 4.0 | 4.2 | 4.4 | 4.5 |
Danish Krone | DKK | 1.4 | 1.6 | 1.8 | 1.9 |
Hungarian Forint | HUF | 4.6 | 4.8 | 5.0 | 5.1 |
Japanese Yen | JPY | 1.7 | 1.9 | 2.1 | 2.2 |
Korean Won | KRW | 3.7 | 3.9 | 4.1 | 4.2 |
New Zealand Dollar | NZD | 3.6 | 3.8 | 4.0 | 4.1 |
Norwegian Krone | NOK | 3.1 | 3.3 | 3.5 | 3.6 |
Polish Zloty | PLN | 3.8 | 4.1 | 4.2 | 4.3 |
Swedish Krona | SEK | 1.8 | 2.0 | 2.2 | 2.3 |
Swiss Franc | CHF | 1.4 | 1.6 | 1.8 | 1.9 |
UK Pound | GBP | 2.4 | 2.7 | 2.8 | 2.9 |
US Dollar | USD | 3.0 | 3.3 | 3.4 | 3.5 |
Euro | EUR | 1.3 | 1.5 | 1.7 | 1.8 |
Notes
1. The DDR is subject to annual change on the 15th January. 2. The formula is as follows :
DDR = Average of the CIRR + Margin The values of the Margins are set out in Article 40 of the Arrangement.
The average of the CIRR for all currencies is calculated taking an average of the monthly CIRRs valid during the six-month period between the 15th August of the previous year and the 14th February of the current year.
Annex 9 Calculations of risk spread from upfront premium fees in EXIM bank loan: 10-maturity, equal repayment of principal
EXIM Bank risk category | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
Upfront premium fee | 1.32% | 2.50% | 4.13% | 6.37% | 9.05% | 11.60% | 14.92% |
Implied annual premium* [1] | 0.25% | 0.48% | 0.79% | 1.22% | 1.73% | 2.22% | 2.85% |
Interest rate (US$ CIRR) [2] | 2.70% | 2.70% | 2.70% | 2.70% | 2.70% | 2.70% | 2.70% |
Total annual charge [1]+[2] | 2.95% | 3.18% | 3.49% | 3.92% | 4.43% | 4.92% | 5.55% |
PDR = 1.70%
CIRR = 2.70%
* Charged on the outstanding loan balance from origination and using a Premium Discount Rate (PDR) after the Zero point.
Source notes and references
[1] https://www.oecd.org/dac/financing-sustainable-development/modernisation-dac-statistical-system.htm
[6] It is noteworthy that deferred payment of export credit premium fees, which are established from the perspective of the government extending export credits according to WTO rules, are discounted at CIRRs minus the 100 BP margin
[8] https://www.cgdev.org/blog/measuring-oda-four-strange-features-new-dac-debt-relief-rules Euan Ritchie’s article for the Centre for Global Development explores the double-counting and other “strange features” of the debt relief rules in some detail.
[9] Minor point, but this caveat ignores the time value of money. A 10-year loan with a single bullet repayment including accrued interest, forgiven at year 10 DOES involve more donor effort at the time of forgiveness than the original nominal face value of the loan at year 10.
[10] Simple example: loan with interest rate of 4%, where donor cost of funds 1% and credit risk deemed to be 4%, only 75% of the credit risk is being covered. Therefore, it would be justifiable to allow, on a quota share basis, 25% of the costs of debt relief as additional ODA.
[12] Section 1.2, https://www.oecd.org/dac/financing-sustainable-development/development-finance-standards/Reporting-Debt-Relief-In-Grant-Equivalent-System.pdf
[13] Item j of the Illustrative List of Export Subsidies prohibited under the WTO Agreement on Subsidies and Countervailing Measures: “The provision by governments (or special institutions controlled by governments) of export credit guarantee or insurance programmes, of insurance or guarantee programmes against increases in the cost of exported products or of exchange risk programmes, at premium rates which are inadequate to cover the long-term operating costs and losses of the programmes.”
[14] It is worth noting that these experts are from the same governments that comprise the DAC (except that Turkey is not a member of the DAC).
[16] To show the spread, UMICs: two in category 2, seven in category 3, eight in category 4, nine in category 5, nine in category 6 and 13 in category 7. LMICs: four in category 3, one in category 4, seven in category 5, 18 in category 6 and 15 in category 7. See Annex 5.
[17] The calculations of the risk spreads from EXIM Bank’s premium rates have been calculated for the US Dollar. For those currencies that have lower CIRRs (e.g. the Euro and the Japanese Yen), the spreads would be even lower (because the Export Credit Participants have agreed that premium fees paid after the starting point of credit should be discounted at the relevant CIRR minus 100 basis points).
[18] https://www.cgdev.org/publication/mismeasuring-oda-how-risky-actually-are-aid-loans-0 In this paper, Euan Ritchie observes that, based on past data, the implied credit risks may be considerably overstated in the DAC discount rates, further exaggerating the ODA in concessional grants. The analysis of OECD Country Risk Experts supports his view.
[19] This does not however mitigate the over-counting of ODA in loans to these countries, as the full value of the “grant equivalent”, inflated by an unjustifiable 9% discount rate, is counted as ODA – not just the level of the grant element above the 25% hurdle.
[22] https://www.publishwhatyoufund.org/2020/09/debt-relief-as-oda-why-its-looking-bad-for-aid-transparency/ This paper by “Publish What You Fund, The Global Campaign for Aid and Development Transparency” notes the “strong moral case” for not counting write-offs for loans for military equipment etc as ODA.
[23] This has been amended from the original paper, as it has been pointed out that the DAC counts relief of military equipment only as OOF - see para 85 https://one.oecd.org/document/DCD/DAC/STAT(2020)44/FINAL/en/pdf
[30] https://www.oecd.org/dac/financing-sustainable-development/modernisation-dac-statistical-system.htm
[31] It is recognised that the question of minimum thresholds for determining ODA may need to be revisited following adoption of a more robust basis for calculating donor effort.
[32] https://dea.gov.in/sites/default/files/ClimateChangeOEFDReport_0.pdf This “discussion paper” prepared by India’s Ministry of Finance, was one of many such critical reports.
[36] During a review of the CIRRs in the late 1990s, the rationale for the 100 basis point margin could not be definitively established. Some argued that it represented “warehousing costs”, others that it was to cover administration. The most likely conclusion was that the 100 bp was a proxy estimate of the margin between the cost of finance for a government and a first-class borrower (e.g. an established bank). In any case, there seems to be no rationale for including it when setting a discount rate based on donor cost of funds (when the loan is in their own currency).
[38] In fact, under a system where debt relief is counted ONLY at the time such costs are borne to a lender, such a restriction would in inappropriate. A 10-year loan with a single bullet repayment (including accrued interest), forgiven at year 10 DOES involve more donor effort at the time of forgiveness than the original nominal face value of the original loan at year 10.
[41] A simple hypothetical example could be where a donor with cost of funds of 2% extends a loan to a developing country with a credit risk warranting a 4% spread. If the interest rate of the loan is 5% (implying a credit risk spread of 3%), this would not qualify as an ODA loan as it would yield no grant element using a discount rate of 2%. However, in case of debt relief, a donor could argue legitimately that the credit risk had been subsidised. Accordingly, a “quota share41” of the costs of the debt relief equivalent to the portion of the risk not covered by the credit spread in the loan - in this case 25%: (4% - 3%)/4% – could legitimately be considered as additional ODA.
[42] https://www.oecd.org/trade/topics/export-credits/documents/pdr-en.pdf (dated 8 February 2022)
[43] It should be noted that these spreads are “currency sensitive” as the discount rate is based on the CIRR. Although not a huge difference, the spreads are even smaller for the Euro and Yen (currencies of many ODA loans) as the CIRRs are much lower than that of the US Dollar.
[44] https://www.oecd.org/dac/financing-sustainable-development/modernisation-dac-statistical-system.htm
[45] ONLY valid if the policy of debt relief as additional ODA is reversed
Original paper cover
Comments