Steve Cutts: My Meeting with WP Stat and OECD Officials, May 2022
Updated: Oct 23
About this meeting record
Stephen J. Cutts’s personal account of his May 2022 meeting with representatives of the DAC Working Party on Development Finance Statistics (WP-STAT) and OECD Development Co-operation Directorate (DCD) officials. The record covers the exchanges before the meeting, his presentation, the arguments made in response and his counter-arguments.
The discussion addresses the 5% base discount rate, credit risk and debt relief, export credits, and the governance of ODA accounting. The complete original record follows below.
Editorial note: This is Steve’s personal historical account, not agreed OECD minutes. The original wording and non-attribution of arguments to individual meeting participants are retained. The PDF heading gives 21 May 2022, while the narrative refers to a meeting on 19 May; that date discrepancy is preserved for author review.
Read the original paper: https://www.odareform.org/post/credit-where-credit-s-due
Read the preceding letter: https://www.odareform.org/post/2nd-letter-to-dac-chair-march-2022
Meeting with WP Stat and DCD Officials, OECD HQ 21 May 2022
I. Pre-Meeting Exchanges
On 7 April 2022, the Head of the Financing for Sustainable Development Division in the OECD invited me to a meeting at the organisation’s offices in Boulogne-Billancourt on the afternoon of 19 May 2022.
I had agreed to attend the meeting on 19 May (travelling to Paris at my own expense as the OECD was not prepared to make any contribution to my travel or accommodation costs) on the basis of three “understandings”, which were accepted in writing:
That the meeting, being at the technical level, would comprise a dialogue to enhance understanding and clarify differences so as to inform a future written response to my letters and papers (i.e. it would not be in lieu of a formal written response to my letters to the DAC Chair, which was still outstanding)
That the meeting would focus on the core issues I had raised (i.e. the main methodological flaws identified and the reforms I had proposed)
That the DAC Secretariat would send, in advance, “any purely factual errors in my analysis”.
In the event, the DAC Secretariat did send me a paper in advance of the meeting which was entitled “Factual errors and omissions in the analysis, as identified by DCD Secretariat, April 2022”. However, this made little attempt to identify factual errors, as requested, or even to seriously address the major problems I had flagged. Instead it made claims of bias, partiality or missing perspectives in my analysis. Below are its three headline points (the first with three sub-points), followed by my observations on each:
Point 1. The analysis is partial and the conclusions on the credibility of ODA thus biased.
1.a. Criticisms (‘flaws’) are not put in perspective, the positive aspects of introducing the grant equivalent system with discount rates differentiated by income groups are missing
Not only was this not pointing out any factual error in my analysis, it was in itself a factually incorrect statement. As my paper pointed out, I had argued strongly for a grant equivalent approach to measuring aid for many years before the DAC adopted it. But that did not mean I could support a set of parameters for calculating grant equivalents that grossly inflated the real level of aid.
1.b. The measure and quantification of concessionality is not as straightforward as presented in the analysis. The DDRs are not recognised as the universal benchmark for measuring concessionality as the note suggests. It is therefore a matter of interpretation that ODA figures would be “inflated” against this benchmark.
This is also not pointing out any factual errors in my analysis. Moreover, it again misrepresents my analysis. At no time did I claim that the Differentiated Discount Rates are a universal benchmark – indeed, I did not even advocate for them to be used for the calculation of ODA. However, it was hardly just “a matter of interpretation that ODA figures would be ‘inflated’ against this benchmark”. For it was simply a mathematical fact that the DAC discount rates had been consistently far above the DDRs, and had therefore generated inflated ODA grant equivalents compared with what DDRs would have produced.
1.c. The note also omits to state that: i) the new discount rates (between 6 and 9 per cent) are lower than the discount rate applicable before the reform (10 per cent). Concessionality rules are now therefore generally stricter than before.
Again, this point is only a claimed omission, not a factual error in my analysis. Moreover, the claim is not even correct, since my paper had explicitly noted that the DAC’s new discount rates were lower than under the previous cashflow methodology.
The Secretariat’s conclusion that “concessionality rules are therefore stricter than before” also requires heavy qualification. Under the old cashflow methodology, the discount rate was only applied as a criterion for a loan to qualify as ODA. It did NOT determine how much ODA was scored, it only determined which loans would have their cash flows scored as ODA. But if the loan was scored as ODA, its repayments had to be deducted from ODA, so that in the end net ODA was zero on every loan anyway. This means that, in practice, the new concessionality rules are actually FAR more generous than the old method, since they score positive grant equivalents for every single ODA loan upfront, and never deduct any repayments. Hiding that fact is an important omission in the Secretariat’s own note, which leaves the reader with a completely false impression of the impact of the new system.
Point 2. The note blames the DAC method for ‘encouraging OECD donors to feed an impending debt crisis through incentivizing loans instead of grants’ but omits to mention the safeguard in the ODA definition regarding debt sustainability (which did not exist before the reform).
Again, this is not a factual error in my paper, as the issue was not mentioned. Moreover, the DAC Secretariat is exaggerating the importance of this safeguard: it only applies to countries already under IMF/World Bank debt management. It does nothing to stop loans to other borrowing countries. Even to countries under debt management, loans can still be given if they meet the WB/IMF concessionality test…and the new DAC rules encourage such lending, even to these countries, rather than grants.
Point 3. DAC members’ ODA and grant equivalent system [sic] does not explain the large share of loans in overall climate finance.
This is also not a factual error, because it is not what I said. My points were that the system encourages the use of loans rather than grants in climate finance, and misrepresents the donor effort in the loans. Both of these points are true, and not contradicted in the Secretariat’s note. Whether the system is already changing the composition of climate finance - the point on which the note takes issue - is as yet hard to determine, as one would need to see several more years of data to confidently identify trends.
II. The Meeting Itself
Having prepared my presentation in advance, and taken notes during and immediately after the meeting, I can present a detailed and I believe accurate record of this meeting. As we agreed that the Chatham House Rule would apply, I have not attributed any arguments to any individual participants other than myself. And, as can be seen, my arguments, which are set out below, only reflected and reiterated my established and public positions.
Present were:
Member State representatives: Jérôme Le Roy (EU, WP-STAT Chair), Megumi Muto (Japan, WP-STAT Vice-Chair), Miryam Rordorf-Duvaux (Switzerland, WP-STAT Vice-Chair), Maher Mamhikoff (Canada, former WP-STAT Chair), Virtually: Thomas Garreau (Trésor, France, WP-STAT member), Katrine Heggedal (next chair of WP-STAT)
Secretariat (included): Senior Counsellor and Head of Division, Financing for Sustainable Development Division; Head of Unit; Development Finance Analyst and Development Finance Analyst
The meeting started around 25 minutes late as I had been sent an invitation to the OECD Offices in Boulogne-Billancourt. The meeting venue had been switched to the Château de la Muette (OECD’s Headquarters) without my being told.
i) My presentation
At the outset of the meeting, I set out three major flaws in the reformed ODA counting methodology:
The universally applied base rate of 5%, which massively exceeds the funding cost of DAC members extending ODA loans.
The addition of credit risk margins to this base discount rate. The 2014 papers sent to me by the Secretariat ahead of the meeting had repeatedly recognised that adding these margins meant that it would not be credible to count any further ODA if debt relief were subsequently given on these loans. Yet in 2018 and 2020, the DAC decided to keep reporting additional ODA at the time debt relief is given, thereby effectively double-counting the credit risk.
The counting of ODA for the cost of debt relief for non-concessional loans, especially export credits. This typically entails no donor effort or cost to the taxpayer, as risk-related premiums have been levied already to cover long term operating costs and losses.
I noted that the flawed reforms reveal a significant governance problem. DAC members are yielding to political pressures to exaggerate their own generosity, and sound statistical principles have been discarded in an effort to overcount ODA.
I also pointed out that there were many errors and omissions in the Secretariat papers referred to me in advance of this meeting. Their overall effect was to hide even from DAC members the real impact of the reform, in terms of its exaggeration of ODA over the costs incurred by the donors. Furthermore, they did not at all substantiate the Secretariat’s claims that the negotiations had been transparent. In particular, while there was reference to the Lowcock Group, no records were shared about its discussions and considerations even though this was the group that effectively brokered the agreed reform on ODA loans in 2014. On the other hand, the documents made clear that in some cases, members had been strongly advised, in the interests of statistical integrity, to avoid decisions that they later took anyway.
ii) Discussion
The following is organised issue by issue. First, I set out each of the three basic points on methodology I made, plus my point on governance. Then I specify the relevant argument(s) made in response by a member or members of the Working Party or Secretariat, each of which is followed by the counter-arguments I then presented.
A. The Base Rate of 5%
My point: The universally applied base rate of 5%, which is much higher funding rate of DAC members extending ODA loans (as evidenced by their long-term government bond yields), means that donors score ODA at levels far above the budgetary cost of extending such loans.
WPStat Argument:
1. The discount rates applying to all donors are simple and easy to apply and rooted in the international system as it aligns the DAC with the IMF. Other options were considered, including the DDRs, but these were deemed too complex, and impracticable as they change too frequently (annually).
My counter-argument:
The DAC may have considered other options, but it settled on one that bears no relation to donor effort. The IMF’s 5% rate was based on CIRRs for the US Dollar – a currency hardly ever used for ODA loans – dating back to 2002 so is outdated and irrelevant. Moreover, the IMF had very specific institutional reasons for moving to a flat 5% discount rate, including that it provides lending in many different currencies (unlike DAC donors) and in currencies where neither CIRRs nor DDRs are published (again, unlike DAC donors).
For DAC statistical purposes, such a discount rate is hopelessly outdated, stagnant and has the effect of massively exaggerating the donor effort in the scoring of ODA.
DDRs are also simple to apply – the relevant discount rate can be obtained in seconds. And the claim that setting discount rates on an annual basis is “too frequent” is not credible. Financing institutions are flexible and adaptable, constantly changing the terms and conditions of their loans to reflect changes in markets, financial conditions and other external factors and circumstances. The DDRs change once each year…and the fact that DDRs change during the tenure of a loan does not change the ODA – it is fixed from disbursement.
WPStat Argument:
2. The discount rate is appropriate as it has not resulted in an increase of ODA. Therefore, while the 5% might look arbitrary, it has worked quite well. And it’s a fair measure of donor effort.
My counter-argument:
Comparing total ODA figures under the new methodology with those under the cashflow system makes no sense. If the grant equivalent system is to be deemed robust merely because it yields the same results as the previous cashflow system, why move away from the previous system?
More fundamentally, a grant equivalent system, even if calculated using far lower discount rates equating to the lender’s costs in extending loans (and not double counting credit risk), will always show higher ODA over time. This is simply because, under the cashflow methodology, every single loan, once repaid, yields net zero capital-flow ODA, whereas under the new system a positive grant equivalent is recorded in perpetuity for every concessional loan.
Thus, the argument that the DAC’s new methodology, by using excessive discount rates and double-counting the credit risk, exaggerates the ODA in loans does not rest on needing to show that overall ODA has increased in the few years since the reform, since the two systems have different temporal effects. In fact, one might have expected to see lower grant equivalent figures in the short term, as many of the lending programmes of DAC donors were still relatively immature (in other words they were still at the stage where the outflow on new loans far exceeded the reverse flow of repayments of older loans). The fact that this did not happen is further prima facie evidence that the discount rates the DAC is using are too high, and certainly not confirmation of the validity of the new system.
The statement that the 5% base discount rate represents a “fair measure of donor effort” is baseless: it massively exaggerates donors’ ODA, and no evidence has been presented that shows otherwise.
WPStat Argument:
3. The 45% threshold for ODA for LDCs requires lenders to be extraordinarily generous: “you practically have to give the money away”.
My counter-argument:
This is only true to a very limited extent. In fact, the 45% threshold for LDCs is largely offset by the much higher discount rate for LDCs/LICs (9% instead of 6% for UMICs or 7% for LMICs) which reduces the “effort” a donor needs to make in softening the loan terms to clear the 45% “hurdle”. Indeed, for LMICs, the higher minimum GE threshold (15%) over UMICS (10%) can be MORE than overcome by the higher discount rate: so it is possible for a lender to provide harder terms in a loan to an LMIC than a UMIC and still clear the minimum concessionality hurdle.
WPStat Argument:
4. It was necessary to remove the disincentive to provide loans that existed under the cashflow methodology.
My counter-argument:
It is true that the previous cashflow method scored zero for every loan, once it was fully repaid, regardless of the generosity of its terms. Accordingly, that system undervalued donor effort by 100% for fully repaid loans. However, the DAC’s “grant equivalent” system, because of the excessive discount rates being used, massively over-corrects this so that loans are being over-valued for some donor-lenders by 800% or more. Despite the pledge made by the DAC when it announced the reform, loans and grants are NOT comparable under the new system – a donor scores far more ODA for each Euro or Dollar of donor effort by extending a loan than by giving a grant. This provides a major budgetary incentive for donors to provide loans rather than grants.
B. Credit Risk
My point: The risk of non-repayment is being double-counted as ODA, first through the inclusion of “generous” risk margins in the discount rate, and then again through the scoring of ODA to account for the costs of debt relief when it is agreed.
WPStat Argument:
1. The Paris Club wants debt forgiveness and rescheduling of concessional loans to count as ODA as this incentivises creditors to agree to such debt relief. And this should be seen as a second transaction: the original loan is for a specific developmental purpose, say to construct infrastructure, whereas the debt relief transaction is designed for macro-economic benefit.
My counter-argument:
By including a risk margin in the discount rate used to calculate the “grant equivalent” of the original loan, a donor has already claimed ODA upfront equivalent to the cost of insuring each loan against default, i.e. it has claimed extra ODA to cover the estimated credit risk of that loan not being repaid in full and of the additional costs that it would then face. Thus, the donor has already implicitly assumed in its reporting of EVERY original loan that some loans will not be repaid in full and has “banked” ODA in advance for the additional concessionality that some loans will acquire through relief (without knowing, of course, which loans will require relief). This is the whole point of including a risk margin in the discount rate used to determine the ODA grant equivalent of each loan.
It is also why, when the DAC decided in 2014 to include risk margins in its discount rates (against the advice of a majority of members of its own Expert Reference Group that preceded the Lowcock Group), it gave an undertaking to stop counting the costs of debt relief on ODA loans as additional ODA:
“Given that the new system would value upfront the risk of default on ODA loans, the eventual forgiveness of these loans would no longer be reportable as a new aid effort.”
The decision to count, as a new aid effort, the “additional concessionality” involved in forgiving a defaulting ODA loan is a clear breach of this commitment.
To illustrate the point, one can use a simple analogy of a property insurer that has a portfolio of 100 houses dotted around the country. Let’s say that by the end of the year, the insurer expects to have claims amounting to no more than 2% of the total value of the property insured. On that basis, the insurance company can reasonably charge a 2% premium on each policy to cover the overall risk and ensure its own commercial viability.
What the insurance company cannot then do is to visit a householder the morning after their house burns down and claim that overnight the risk increased because their house caught fire, and that therefore the householder has to pay an additional insurance premium to cover the other 98% of the value of the property.
Yet this is what the DAC is doing in claiming ODA for the “additional concessionality” (i.e. the value of relief) associated with a loan default. The donor has already booked ODA credit for effectively self-insuring its whole portfolio of loans, so it cannot claim additional ODA just for those loans that are not repaid in full. That is clearly “double-counting” the risk.
Looked at another way, donors could only justify scoring extra ODA for the “additional concessionality” in relief if they were also to score negative ODA for all the loans that were repaid without problems and where the extra ODA scored upfront for risk therefore turned out to be unwarranted….but in that case it would be a far neater solution to just score ODA for relief when it occurred and remove the credit margin in the upfront discount rate used to calculate the ODA grant equivalent of every loan.
WPStat Argument:
2. The safeguards are such that we can be assured that we are not double-counting the credit risk.
My counter-arguments:
The ceilings on counting additional ODA credit for relief that are referred to here as “the safeguards” do not eliminate the double-counting: in the analogy of the property insurer above, they just cap the amount demanded after the fire to the full original value of the house minus premia previously paid. So for any loan requiring debt relief, the donor ends up counting the FULL costs of that loan as ODA. Meanwhile, for all the loans for which debt relief is NOT needed, the donor scores completely unwarranted ODA through the credit risk margins included in the DAC’s discount rates.
WPStat Argument:
3. Many ODA loans have very long tenures, which therefore warrant higher discount rates.
My counter-arguments:
Longer terms do add to uncertainty, but the explanation given (see above) about why there is no justification for including credit margins in the discount rates used to calculate ODA apply regardless of the maturity of the loan.
Moreover, while it is usually the case (though not always) that governments’ cost of funds increases for longer bond maturities (i.e. the yield on a 10-year government bond tends to be higher than for the equivalent a 5-year bond), the yield curve tends to be very flat when looking at bond yields at longer maturities.
Notably, the DDRs do include additional margins for longer maturities, ranging from 75 basis points for loans of less than 15 years up to 125 basis points for loans with maturities exceeding 30 years. These are extremely generous compared with market assessment of long-term risk, but a modest margin for ODA loans with very long maturities for which appropriate bond yields in that currency are not available, may be appropriate….but only if they were being added to realistic base discount rates, and not to an already excessive 5%.
Of course, one might assume that given the nature of its work, the development community might take a more optimistic view and expect to have a positive impact over the long term, thereby reducing the risks of defaults and the need for debt relief as developing countries gain stronger and more resilient economies.
C. Counting ODA for Debt Relief for Export Credits
My point: Only real donor effort could justify scoring grant equivalent ODA for relief of non-ODA loans. But there is no donor effort involved in debt relief for export credit debt (or similar loans) where risk-related premium fees or loan margins have been levied from developing country borrowers (or commercial exporters) to cover those same costs, as is required under WTO rules.
WPStat Argument:
1. Debt relief for export credits has not been part of the work programme in recent years and could be reconsidered in conjunction with the Paris Club.
My response:
The issue clearly needs to be re-opened urgently. The body that (currently) determines what qualifies as ODA and how it is counted is the DAC, not the Paris Club.
If the DAC has not considered the issue, the DAC Secretariat should not have assumed and then asserted in its 2020 instructions for the reporting of debt relief on loans originally classified as OOF, export credits or private flows at market terms that “as no grant equivalents will have been recorded for the original claims, the donor effort involved in concessional debt relief…should give rise to ODA grant equivalents”. The instructions completely ignored the fact that debt forgiveness for most of such loans entails no real donor effort, as the risk of default will have been built in to the pricing of the loan and/or the premia paid to insure or guarantee the repayments.
2. The Paris Club is expected to deal with debts from ODA loans, export credit loans and OOF. The package could cover all sorts of different loans, concessional and otherwise. So it gets quite muddy and difficult to separate out the different loans – premiums may have been paid for some of the loans but not others.
Creditor countries are fully aware of the breakdown of the debt they are forgiving or rescheduling and quite capable of ensuring that they claim ODA only for the ODA and other loans extended at concessional rates where risk fee or spread has not been included in the terms.
D. Governance Issues
My point: The flawed reforms reveal a significant governance problem. Unrestrained by the presence of aid recipients and unheeding of financial expertise, the DAC as a “donors club” has yielded to political pressure and reached a self-serving consensus that flouts sound statistical principles to exaggerate to their taxpayers their largesse and generosity to developing countries.
WPStat Argument:
1. This is not a correct characterisation of the DAC. Some other international organisations and delegates from developing countries are observers to the Committee.
My counter-argument:
What matters is who are the decision makers – not with whom WP-STAT (or indeed the DAC itself) has a dialogue.
WPStat Argument:
2. Every other statistical measure is managed by those who make the effort; on that basis, ODA accounting should remain with the DAC.
My counter-argument
In fact, in line with international statistical principles in the UN and the OECD’s own Council-endorsed Recommendation on Good Statistical Practice, most countries have made their national statistical offices independent of line departments so as to provide them with the independence recommended by these guidelines. Accordingly, the DAC’s clinging onto its political stranglehold over the accounting of ODA is an anachronism that does not comply with the OECD’s own statistical guidelines.
Original document image
Comments