A note on current problems with ODA as a statistical measure
By Simon Scott
10 October 2019
In 1969, the OECD’s Development Assistance Committee (DAC) created official development assistance (ODA) as a measure of foreign aid effort. To qualify as ODA, transactions had to be “concessional in character,” i.e., to give something of value away. In 1970 the U.N. set a target for ODA of 0.7 per cent of donors’ national income.
The DAC has recently changed ODA reporting rules to include transactions that require no financial sacrifice. This deprives ODA of its meaning as a gauge of aid effort, and vitiates the point of setting the U.N. ODA target.
The changes have also rendered ODA incoherent as a statistical measure, making it a faulty tool for monitoring and analysis. ODA now fails to meet basic statistical quality standards.
ODA was never perfect, and for years critics complained about the inclusion of items – such as the costs of students and refugees in donor countries – which transferred no resources to developing countries. But since 2014, DAC decisions mean that ODA is not just counting debatable items, but inventing numbers that do not exist in the real world.
One example is loans to governments and other public sector borrowers. These are now measured not in observed flows of capital but in claimed “grant equivalents.” The grant equivalent is a model estimate of the amount being given away in a loan. It represents the loan’s value, minus the present value of its expected repayments.
Grant equivalents are only credible if donors work out the present value of repayments using current market interest rates. Yet in 2014 the DAC decided instead to use fixed discount rates of 6, 7, or 9 per cent (comprising a “base rate” of 5 per cent, plus a “risk margin” of 1, 2, or 4 per cent, depending on the borrowing country’s per capita income). These high rates underestimate the present value of loan repayments, leaving inflated grant equivalents, even for loans that are actually at market terms.
The DAC’s discount rates also take no account of a loan’s currency or duration, or of the actual country risk of the borrower. So while the DAC’s grant equivalents are systematically wrong, they are wrong by different margins for different donors and loan profiles. Thus they cannot be corrected by any standardized adjustment.
The other fictional figures now in ODA are for equity investments sold at a profit. Here the DAC decided in 2018 to hide the profit, and only report sale proceeds – which generate negative ODA entries – up to the amount of the initial investment, no matter how long ago this was or how much inflation there has been in the meantime. This again biases ODA figures high.
Apart from including fictional figures for loans and equities, ODA has now become incoherent as a statistical quantity. In 2014 and 2016 respectively, the DAC had recognized that introducing grant equivalent methodology would require new rules on counting debt forgiveness, and loans to the private sector. But by 2018 it had failed to agree on these. Logically it should then have put off introducing the grant equivalent system. But instead it decided to keep on reporting these two items on the old net flow basis – so the headline ODA figure is now a mishmash of grant equivalents and flows.
These fundamental flaws of overcounting, incoherence and premature implementation of an unfinished system ramify through the ODA reporting rules, generating anomalies and inconsistencies. For example:
- Donors making the same loan to the same country in the same year record its full amount if it is to the private sector, but only its “grant equivalent” if it is to the public sector.
- They report the loan’s repayments if it is to the private sector, but not if it is to the public sector.
- There are no rules on whether to report repayments if a borrower changes from public to private or vice-versa through privatization, nationalization, or the invocation of a guarantee.
- Donors report identical amounts for forgiving a loan to the private or public sectors, even though they will have reported different amounts up to that time for the same loan performance (net flows for private loans; grant equivalents for public loans).
- The amount donors report for forgiving a loan to the public sector will depend on how much the borrower has repaid, even though donors will not have reported these repayments in the grant equivalent measure.
- To calculate grant equivalents, donors use discount rates that include the risk that they may have to forgive the loan, but if they do, they will report the amount forgiven as well, thus double-counting risk, contrary to the DAC’s 2014 undertakings.
- Donors may report loans to the private sector individually, or as subsidies to loan-extending agencies, or not at all. Their figures are thus not comparable, and the DAC total for this item is meaningless.
- Headline ODA figures are now a broken series, with data before 2018 being on the flow basis, and data from 2018 being on the hybrid flow-and-grant-equivalent basis.
- From 2018 the DAC only publicizes ODA/GNI ratios on the new hybrid basis, while the official U.N. ODA/GNI targets, including for Sustainable Development Goal indicator 17.2.1, are for actual net flows.
- The DAC’s decisions on its “grant equivalent” method also inflate ODA flow figures: loans that are not concessional in character, but for which grant equivalents are recorded, count as ODA under the flow basis as well; and the flow basis also ignores profits on equity sales.
- Donors showing the same ODA totals may have generated them either from genuine aid spending, or from market-terms or mispriced transactions. This means that ODA is now an unreliable metric for comparing aid between donors and over time, towards regions and income-groups, by sectors, and in terms of its concessionality or degree of tying.
- The ODA definition, which until 2014 comprised about 80 words, is now almost 800 words long. It includes nine footnotes, mainly to explain the “status of DAC discussions” on still undetermined issues, or the need to review sections following the 2016 High Level Meeting – which took place more than three and a half years ago. It, and most of the associated new ODA reporting instructions, will be incomprehensible to the public, and confusing even for experts.
More broadly, the new “headline” ODA figure is incommensurable with the other DAC aggregates of official and private finance, all of which are presented on a flow basis. It is also inconsistent with the proposed new measure of Total Official Support for Sustainable Development (TOSSD), which distinguishes between concessional and non-concessional flows using a far tougher test (35 per cent grant element at a 5 per cent discount rate) than for ODA grant equivalents. This means that ODA will include as concessional loans that TOSSD classifies as “non-concessional.”
The bottom line is that ODA can no longer fulfil its function as an objective measure of donor aid effort. Its rules exaggerate donor effort, and its total mixes different statistical quantities. The resulting data lack credibility and usefulness. ODA’s definition, coverage, and reporting instructions all need urgent and fundamental review.
This piece was originally published by the Brookings Institution.
About the author/s
Simon Scott is former head, DAC statistics division, OECD