Hedwig Riegler: Governance of ODA accounting – no substantial ODA reform without governance reform

About this paper

In this April 2023 guest contribution, Hedwig Riegler sets out 11 proposals for reforming aid and resource-flow statistics. Drawing on her experience as Austria’s DAC statistical reporter and as Chair and Vice-Chair of the OECD/DAC Working Party on Development Finance Statistics, she considers both the technical rules and the institutions that set them.

The paper argues for clearer definitions of donor effort, a distinct place for non-ODA finance, better measurement of real transfers and statistical decision-making independent of political influence. It also examines grant equivalents, debt relief, guarantees, Country Programmable Aid, climate finance and TOSSD.

Editorial note: This is Hedwig Riegler’s personal view from April 2023. Her judgments, proposals and references to the rules, institutions and data available at that time are retained in their original historical context.

Riegler on new accounting system_final.pdf

Download PDF • 285KB

Read the preceding analysis of ODA accounting: https://www.odareform.org/post/point-by-point-repudiation-of-the-oecd-faq-defending-oda-accounting

How to reform aid and resource flow statistics

A personal view by Hedwig Riegler, April 2023

A long and still growing list of critiques on ODA accounting provides evidence of the DAC’s rule-changing spree of recent years. The DAC has succumbed to pressures, especially from finance ministries, to agree ever more generous ODA rules in order to make its members' performances look much better than they are. The process has become almost pathological, and has damaged the DAC's ODA concept and methodology beyond repair. A fresh start is needed to restore credibility to aid and resource-flow statistics. In this piece, Hedi Riegler, a long-standing DAC statistical reporter and former Chair and Vice-Chair of its statistical working party, offers some personal reflections on what a sound new statistical architecture for ODA and other flows would look like, and how it should be developed.

As the rest of this website shows in detail, “ODA reform” is urgently necessary. But what would a new, credible system look like, both institutionally and technically? As Austria’s DAC statistical reporter for many years, including service as Chair and Vice-Chair of the DAC’s Working Party on Development Finance Statistics (WP-STAT), I have been reflecting on this question in the light of all that has gone wrong with ODA and related measures in recent years. I offer the following initial thoughts on what might be the basic steps in developing an effective, reformed statistical system for measuring aid and other resource flows for development.

1) Go back to the roots

The original basic set-up of OECD/DAC resource flow statistics, which since the 1960s has aimed at collecting data on all financial flows relevant for development, was the right decision. To define ODA (Official Development Assistance) – official concessional finance – as a subset thereof was right, too.

However, it was a mistake to then focus solely on ODA – treating it as more or less the only finance for development, and neglecting all other flows both in statistics and policy discussions. This laid the foundation for the trend of seeking to squeeze everything into ODA, whether it fitted or not, and to undermine the special role of ODA within the landscape of overall development finance.

ODA should be targeted to regions, sectors, or projects that have little or no access to market financing, while market or market-near instruments should be based on open competition and the play of free market forces. As demanded by developing countries, ODA was defined as a separate category of official and concessional flows with its own target level in order for recipients to have predictable, plannable resources at hand for sectors or undertakings that could not attract financing at market terms.

Private sector finance, a new focus in development co-operation since 2011 (Busan 4th High Level Forum on Aid Effectiveness) can contribute a lot to development, if a proper framework is in place (which ODA finance can help to build), but it has a different nature; private sector finance and ODA should always be treated as complementary yet distinct financing streams. What has gone wrong is that the borderline between ODA and market-based financing for development has become completely blurred. The DAC has been trying to jam “private sector instruments” (PSIs, i.e. instruments usually offered at market terms) into ODA, using questionable and misapplied methodologies. It cleared the way for this by replacing ODA’s concessionality requirement by an alternative “additionality” criterion after having admitted that PSIs were non-concessional in nature. It is now trying to reach formal agreements on incorporating various PSIs into ODA on this basis. This will demolish a definitional pillar of ODA, and it will only be a matter of time before the whole ODA concept crumbles.

Main categories to be collected

A reformed system would need to re-establish the traditional and correct understanding of the universe of financial flows for development. It would respect the fundamentally different natures of the components of this universe, while paying renewed and proper attention to the non-ODA flows, and, if necessary, adapting the main- or sub-categories to new requirements:

a. Official concessional finance (ODA – Official Development Assistance)

b. Official non-concessional finance (OOF – Other Official Flows, OEC – Official Export Credits)

c. Private concessional finance (aid from NGOs, foundations, etc.)

d. Private finance at market terms (FDI etc.)

All the items in this list are flows of resources on which the DAC has collected data since the 1960s. They can be added together to form the traditional picture of total resource flows for development, whether expressed in terms of gross or net disbursements. Items a. and b. can also be expressed in terms of commitments (i.e. firm contractual agreements).

ODA, first on this list and the quantitative synonym of “development co-operation”, has become a mix of grant flows and a growing share of non-grant flows/concessional loans with grant equivalents. Controversy has arisen as to whether flow and grant equivalent data can be added together. In my opinion they can, because it is not really the flow that is being counted on a grant, but rather the grant equivalent of that flow (which is 100% for a grant).

However, I have a “narrow” understanding of the applicability of grant equivalents and think they should only be calculated where there is or is likely to be an official flow that would be recorded in the above system. By contrast, it seems that the predominant thinking has become that grants represent cash-flow, whereas grant equivalents are the opposite: non-cash-flow. Such thinking holds, for example, that the measure of donor effort can now be expanded to non-flow financial instruments (e.g. guarantees), including instruments that stimulate private flows, but which the official sector offers at market or near-market terms.

I have a different understanding: grant equivalents to me are anticipated implied official grants that, in real life, simply happen at a different – often much later – date than the extension of the finance instrument: e.g. subsidies to reduce interest charges on loans are liable to be paid out over the life of the loan, in step with repayment schedules. Such grants normally are in-donor-country transfers which result in a direct financial benefit for the recipient country (e.g. an interest-free or near-interest-free loan). It makes sense to count grant equivalents that represent a time-delayed actual grant at the point of extending the long-term financial flow, though only if grant equivalents are approximated closely to the anticipated real grant. Such grant equivalents are therefore appropriate – and should be applied – for ODA loans, using market-based discount rates that accurately estimate a “gift share” which stands for a real financial sacrifice – an anticipated fiscal loss – over the life of the loan.

Grant equivalent accounting, however, is in my view not appropriate for many other long-term instruments: not for non-flows, and definitely not for PSIs (private sector instruments) extended by DFIs (Development Finance Institutions), which usually operate on a financially self-sustaining basis with no anticipated financial losses that will require taxpayer compensation and which instead even make a profit. Calculating grant equivalents on PSIs represents a misuse of the procedure, applying unrealistic discount rates to transactions on market terms – including guarantees and lines of credit that may not involve any official spending – so as to arrive at a virtual grant that is never given in real life. In these circumstances, even the term “grant equivalent” is inappropriate. Advocates of this line of thinking seem to be thinking of “donor effort” as some sort of virtual reality that they can call into being and record as fact, even in the absence of any budgetary spending. I do not agree with this expansive notion of “donor effort”, and would remind its proponents that “real budgetary spending” was the original intention of ODA – which to date has not been changed by its principal, the United Nations.

This is not to deny that official support for development may include non-flow items, such as guarantees, other contingent liabilities, tax deductions for donations to developmental NGOs, or tariff concessions on developing country imports. It may be worthwhile considering these measures, and it may be feasible to obtain and present information on them, and possible to give donors credit, as appropriate, for any financial or other effort involved, but this information should be clearly distinguished from the flow figures mentioned above.

Flow figures are the original, authentic material needed to make sound statistics. DAC’s statistical collection on development finance is descriptive statistics, which means it is based on observed facts, not on probabilities and not on complex modelling with a multitude of assumptions. A new system should keep it that way – or rather restore it that way: as, first and foremost, a source of data.

2) Clarify the WHAT of measurement

For ODA, the long-standing interpretation by most DAC statistical reporters has been that it is a measure of donor effort. However, this very basic definition required for any statistical collection (I call it the “underlying rationale of measurement”, i.e. the logic to be applied to practically all major decisions on what to measure and how to measure it) was never laid down explicitly in the DAC Statistical Reporting Directives until the 2015 edition. And even then it was just baldly stated in a single phrase, which related only to the newly introduced grant equivalent measure, as if this had never applied to cash-flow measurement (“The ODA grant equivalent is a measure of donor effort.” Para. 58 in DCD/DAC/STAT(2015)31). This definitional shortcoming led to confusion and dispute in WP-STAT debates, in particular on the concessionality of ODA loans, some members claiming that ODA measured recipient benefit, even though this is a quite different concept from donor effort. An answer to this very fundamental question in statistical measurement is key, because it not only determines methodological approaches, from the point of measurement to the choice of discount rates in grant-element calculations, but it also impacts heavily on the resulting data. Changing the rationale of measurement within a system (e.g. ODA) inevitably leads to inconsistent data and meaningless aggregates of “apples and pears”. The DAC’s approach to measuring ODA in private sector instruments is doing precisely that: switching back and forth between a “donor effort measurement” (concessionality) and a “recipient benefit/impact rationale” (financial additionality/developmental additionality).

This ambiguity in the current system must be closed in a new accounting system and a full, accurate, and rigorous statement must set out the underlying rationale of measurement. The rationale must be rigorously defined and applied both at the level of the four main categories of flows mentioned above and at the level of any more detailed measures at different points of measurement.

3) Enhance point of measurement awareness

For both the reporter and user communities, increased awareness and better knowledge about the implications of points of measurement chosen for a statistical data collection would vastly improve both data presentation and use.

Currently, aid is measured separately at two points: 1) at donor outflow point and 2) at recipient country inflow point. Based on these two separate measurements, statistics are produced by the OECD/DAC Secretariat on 1) donor provision of ODA (counting towards the 0.7 % target) and 2) aid inflows to recipient countries, which includes concessional inflows from multilaterals and large non-governmental organisations (e.g. Bill & Melinda Gates Foundation). These statistics provide the basis for production not only of the useful donor and recipient aid charts available online, but also of a wide range of statistical products that feed into policy discussions.

Measurement at these two points should by all means be retained, but the system needs to be refined and extended. One definite refinement should be clearer distinction between official concessional vs private concessional inflows to recipient countries (e.g. multilateral vs Gates Foundation), and one additional measurement should be that of “real-transfer aid” [1] at both provider outflow and recipient inflow point, perhaps by reforming the DAC’s existing measure of CPA (Country Programmable Aid), as described below.

4) Make CPA – Country Programmable Aid – a more useful measure

CPA can already be considered as a sort of approximation of real-transfer aid; however, the decision to use it as a measure of planned future aid spending subject to decision-making by both donor and recipient and to embed it in the “Forward Spending Survey” gave it a somewhat different profile and excluded real-transfer aid that cannot be planned (e.g. disaster relief). It may be useful to consider reconfiguring CPA as a measure of total real inflows to each developing country.

The data presented in the reports from the Forward Spending Survey (FSS) implied that CPA is a mixed measure of (planned) donor outflows and recipient receipts. In addition to forward spending figures, tables in the report formerly included actual spending data with a percentage-of-bilateral-ODA column. On this percentage, Austria invariably ranked last among DAC members. The data, albeit just a rough approximation of “core aid”[2] actually transferred to developing countries, were extremely useful in our domestic debate on the skewed, flawed composition of our ODA portfolio.

It seems that publication of the FSS report was suspended after the 2012 edition (on forward spending plans up to 2015), and the online database now only shows old projections from the 2016 Survey that finish in 2019. It thus appears that this Survey has been phased out. This may have been a sensible economy as I expect that FSS data would hardly be missed by those for whom OECD/DAC claimed to be producing them: developing countries. I know from my work as statistical reporter, which included the provision of statistical data for domestic needs, that even in the period of publishing the Surveys, partner countries would request planned spending data from their donor partners directly, in formats defined by them, in advance of bilateral aid programme negotiations. And donor partners would supply these data directly (it was part of my work); developing countries would not go through OECD/DAC to get the planning data they needed, because they found this impractical.

However, CPA data are still calculated and are available through an online query. Therefore, since CPA still exists, and has been freed of its role in underpinning the FSS, the opportunity arises to turn it into a measure of “real-transfer aid”, by adding back in amounts that represent real transfers but which are not programmable in advance (such as humanitarian aid). This would add value to the data set and strengthen the role of the DAC, as a donor forum, in monitoring donor behaviour in extending aid.

5) Separate statistics from politics

Statistics are supposed to be made for not by political actors. Statistics under political influence are prone to be corrupted sooner rather than later. For good reason, both the United Nations Fundamental Principles of Official Statistics and OECD’s own Recommendation on Good Statistical Practice demand that producers of official statistics be professionally independent. Following these recommendations, many countries began to reorganise the institutional set-up of their official statistics by enhancing the co-ordinating role of their national statistical offices and moving them away from direct political control of the government. The organisational set-up of the DAC is an anachronism in this regard as it does not conform to these requirements and developments, and a reform of the current arrangement of members’ political delegates making decisions on statistical methodology (ignoring technical recommendations) is long overdue.

6) Take the development of statistical rules out of the hands of the DAC

Based on my long-term engagement with the DAC regarding statistics, I harbour no hope that positive changes, let alone the fundamental reform required, will be possible from within. It would take a decision by political delegates to remove themselves from the power of statistical decision-making – and that can hardly be expected to happen (at least without massive external pressure).

7) Place aid and resource flow statistics within the OECD as an independent body

Over the years, many calls, including from developing countries (e.g. in ECOSOC meetings), have been made to move ODA statistics away from OECD/DAC and to the UN. Personally, I have never been in favour of this option, because of what I have seen in terms of data collection and presentation by UN organisations. My experience with UN-initiated endeavours in this area stems from my involvement in climate finance reporting to UNFCCC, reporting on the Rio Conventions, UNDP Development Cooperation Reports in developing countries and similar reporting on financing for developing countries. Despite all these initiatives, whenever one queried the UN website for SDG indicators, it invariably connected you to OECD databases for a large number of indicators of developmental flows. Thus, it is difficult to avoid the impression that the UN simply does not have the capacity, experience, or will to do a professional job of collecting and presenting statistics, at least in this field.

By contrast, the OECD seems to possess all three prerequisites: capacity, experience, and the will to take on a commission to produce robust statistics. When I started my job as an OECD/DAC statistical reporter, I was impressed by the professionalism and soundness of the work done at OECD. Not everything was good; for example, the USIF instructions for item-level reporting to the CRS – Creditor Reporting System – were hopeless when I started in January 1996. However, I was impressed by how improvement on this issue was tackled – and I was highly motivated to contribute to this and other processes of developing the statistical system further.

Things began to go drastically wrong when non-aid ministries of member governments sensed that barriers to political interventions were receding and that it was easy for them to influence functions and fora, even, or perhaps especially, technical groups like WP-STAT.

So, whatever the organisational set-up – an independent statistical commission, satellite body, working group, or whatever other form – this must be effectively removed from the influence and control of policy ministries in member governments, especially finance ministries, which have a natural interest in budgetary savings, whereas ODA is (or at least should be) about budgetary spending.

8) Clean up ODA accounting

This is a field that deserves separate special attention, which it has received both in other contributions on this website and through other publication channels; therefore I will limit myself to some general principles and key points that I have rather strong feelings about.

Start with the whole picture of development finance. The discussion on a new system should not focus exclusively on ODA accounting, but should discuss reforming the entire architecture of DAC statistics. At the time of its initiation by OECD/DAC Chair Brian Atwood, TOSSD (then called TOSD) was intended to extend the coverage and increase the visibility of official non-ODA flows in order to offer a more complete picture of official support for development. But later, this initiative was led up a dead-end street, by attempts to turn it into a universal measure. As a result, TOSSD today lacks support both from developing countries, who tend to think of it as a donor club idea and endeavour, and even among many DAC members themselves, who lost interest once it was decided to drop the donor perspective of TOSSD. Therefore, I would be very surprised if this measure ever takes off and becomes established, despite the extravagant claims often made for it. In fact, TOSSD, as currently designed, is in even worse shape than ODA, and requires complete re-engineering.

However, OECD, with the stock of data already collected through its long-established systems, is still the international organisation with the right prerequisites for providing a more complete picture of development finance (including private development finance).

Preferably do not include risk margins upfront when calculating the grant equivalent from the point of view of donor effort of official aid. In the case of loans, if debt goes bad and needs to be relieved, let’s count it at the real cost of default to ODA budgets, if anything. This suggestion is based on my reporting experience for Austria, where any compensation payments for debt relief (e.g. interest reductions on refinancing loans or compensation of losses due to debt forgiveness) would appear as transfers from the central budget to the specialised agency managing the long-term finance instruments. If losses on account of debt forgiveness (e.g. on former export credits) were sufficiently covered by premia collected from private sector actors (these resources are managed in separate, confidential accounts) no transfers from the central budget would be needed. Yet only such transfers would correctly represent real fiscal effort at the point when it is actually made. Therefore, recording them is the preferable method because it ensures the closest approximation to reality, and if, as in Austria, long-term development finance is managed by specialised agencies (and not directly by ministries themselves), necessary compensation payments to cover default or loss are reflected in budgets. It is the same in the case of guarantees: if these are called, let’s again count only the real cost to ODA budgets, and not count anything if the guarantee was purchased with premia that have already compensated the official sector for assuming the risk. That way real official spending on these finance instruments is captured when and where it happens. To try to score risk-assumption upfront across the board not only inflates figures, but also gives false impressions of the timing of actual costs and the nature and magnitude of donors’ real allocations of aid resources. However, if risk margins ARE to be included upfront, the current practice of counting ODA again at the time debt relief is agreed must be stopped: the current overcounting of the credit risk massively exaggerates donor effort and is indefensible.

Purify ODA: Ideally, ODA should exclude expenditures by OECD/DAC member countries primarily related to obligations other than development co-operation. Yet in several cases the DAC effectively allows the reporting of such costs as ODA (applying specific rules). For example, meeting in-donor-country refugee costs and allowing developing country students to access subsidised university places may eventually foster development of the beneficiaries’ home countries, but they are primarily undertaken based on other considerations, whether of domestic policy or to comply with EU legislation or binding international commitments. They would happen regardless of and independently from the international development agenda, and for reasons other than the ODA test of having the “promotion of the economic development and welfare of developing countries” as their “main objective”.

ODA should also exclude operations that involve no spending from official budgets, since this contradicts the definition of ODA as a measure of donor effort. One prominent example of this is the relief of non-ODA debt, where huge volumes can be counted as ODA while not a single dollar is spent from official budgets.

Thus, an effort should be made to analyse the OECD/DAC types of aid and types of finance to identify “candidates” for exclusion.

A purification of ODA would, however, require a rethink of the 0.7% target, which is unrealistically high as it is, but even more so if it were to be met with “core aid” spending only. Questioning the 0.7% target has been a political taboo, but holding on to it is perpetuating the long-standing lie about donors doing everything to reach this target. They repeat this commitment internationally at regular intervals, while at home it is an open secret that many donors are seeking budget-saving ways of reaching the target through some of the “stealth” methods now being encouraged, or at least enabled, by the DAC. Austria is a prime example but it is not alone.

9) Improve visibility of non-ODA flows

A purification of ODA would, at the same time, require giving donors much more credit than at present for non-ODA operations aimed at development. Accordingly, purified ODA, with an adjusted quantitative target, would require a simultaneous upgrade of the status of non-ODA flows for development, so that proper attention is paid to overall resource flows for development.

The wasted effort of starting a TOSSD concept with its own ancillary collection system probably wouldn’t have happened if non-ODA flows hadn’t been treated as a poor relation of ODA. Many decision-makers were not aware that DAC statistics on these flows existed, that an internationally-agreed target of 1% of GNI for all flows (ODA, OOF and private) predated the 0.7% ODA/GNI target, and that the DAC still reports on its members' all-flows/GNI ratios in its standard tables as a broad indicator of their financial cooperation with developing countries, in accordance with UN resolutions. A new accounting system should incorporate the TOSSD objective of providing “a more complete picture” and thus must include the donor outflow perspective.

However, South-South Co-operation should not be included, but instead be left in the hands of others. The UN and South-based organisations are already dealing with quantifying South-South Co-operation and are much better placed to develop their systems further. OECD, with its long-standing reputation as a “donor club”, would never be able to gain the same acceptability as the Southern organisations in this area.

10) Clarify relationship with “adjacent” concepts

Climate Finance: An overlap of ODA and climate finance clearly exists, but has long been denied by many stakeholders, the argument being that climate finance had to be “new and additional” and therefore had nothing to do with ODA. This however ignores the fact that environmental protection and sustainability, including the aims of emission reduction and adaptation to climate change, have been part of the ODA accounting system (and indeed have long had purpose codes for ODA) since long before “climate finance” gained its own quantitative monitoring system. Removal of climate finance data from ODA ex post would not make sense, therefore some overlap is inevitable (although it is not 100%). This issue needs to be analysed in detail and dealt with in reporting instructions so that the delimitations across different accounting systems are clear. The lack of clarity in this area has led to identical activities often being reported in completely different ways, including different financial volumes, to each data collection system, thus creating contradictory and confusing data. As an advisor to Austrian authorities in building a climate finance reporting system, I undertook special efforts towards harmonising Austria’s reporting in this respect, which have been largely successful. But I notice that the approach at international level is still segregation rather than harmonisation.

TOSSD: The case of this newly established measure is similar. The attempt to establish TOSSD as a separate-but-equal conceptual universe from the existing OECD/DAC ODA and other resource flow categories, while at the same time borrowing slabs of text from the OECD/DAC reporting instructions, has led to large, poorly delimited overlaps between the two data sets. Many key subsets in the TOSSD data collection – concessional loans, aid to refugees, peacekeeping expenditures, debt relief etc. – have the same names but different coverage from OECD/DAC data. While, from its inception, TOSSD claimed to be designed as a complementary measure to ODA, in practice it has made a complete mess of this complementarity in mathematical terms. This is the biggest underlying weakness of the TOSSD methodology, which can only be remedied by a re-design from scratch.

Therefore, what needs to be done in a new accounting system for ODA and other resource flows for development is to clarify each system’s relationship with the others through careful delimitations of each set and explicit explanations and quantifications of overlaps.

11) Who should draft a new system?

An independent “group of wise people”, shielded against political interference, could be one option, but if so, the perspective of statistical reporting must be represented. If the group were to comprise experts who have never been involved in practical data reporting, the danger is that rules will not be able to be implemented properly in practice. This has been a perennial problem in the DAC. Just one example: tied aid. In principle, this idea is simple: it means aid money that can only be spent on products from the donor country or a restricted group of countries. But the DAC Untying Recommendation defines tied/untied aid in a vague “political” way which has made proper implementation next to impossible in practical statistical reporting. The abstract definition leaves a lot of room for different interpretations and, therefore, if you look at the tying status reported by members in CRS microdata, you will find out that this is another field of widespread embellishment and misrepresentation. If “political” actors define statistical categories they often (and maybe often inadvertently) resort to diplomatic behaviour and language, which impairs clarity and precision, undermining statistical soundness.

Even professional statisticians who have never been involved in actual statistical reporting are sometimes unaware of potential infeasibilities in implementing statistical definitions in practice. So, the composition of any “wise persons group” to set up a reformed statistical system for measuring resources for development needs to be diverse and carefully considered.

This is just a first input, along very general lines, to the discussion from my perspective as a long-standing practitioner in the field of DAC statistics. I hope that at least parts of it may be useful in due course.

Source notes

[1] Aid that actually crosses the border into the recipient country in the form of finance or personnel engaged in development activities.

[2] Aid from budgets specifically dedicated to development co-operation activities.

Original document image