TOSSD: A Statistical Assessment

stevecutts

Mar 17, 2025

21 min read

Updated: 5 days ago

About this paper

Simon Scott assesses four statistical problems in TOSSD: the fact that no totals are requested or certified, which leads to massive undercounting; data collection and calculation difficulties; conceptual and definitional incoherence of TOSSD as a whole; and the impossibility of determining the value of several TOSSD items as sustainable development finance. The paper explores possible reforms and includes Hedwig Riegler's extended response on statistical governance, coverage and the design of a coherent measurement system.

Editorial note: The original text follows in its historical context. References to rules, proposals, figures and future events reflect the source document, rather than an update to the present day.

Click twice on the link below to download the original pdf:

TOSSD: A statistical assessment

by Simon Scott

with a comment by Hedwig Riegler

Last year I wrote a short blog post on what I saw as a significant problem in the statistical coverage of the Total Official Support to Sustainable Development (TOSSD) measure initiated and still housed at the OECD, though now governed by an International Forum. This article aims to deepen that analysis and assess prospects for improvement.

Almost everyone agrees on the need for a broader measure of international development finance than just Official Development Assistance (ODA) – which basically equates to “government foreign aid”. And indeed the OECD has always offered broader measures, especially of Official Development Financing and Total Resource Flows to developing countries. But these measures are little known and seldom used, and in any case there have been many calls to update all measures to reflect the general concept of sustainable development and/or the specific UN Sustainable Development Goals.

So there is clearly a demand for something like TOSSD. Moreover, its governance – unlike the governance of ODA – is more broadly based than just the donor countries represented on the OECD’s Development Assistance Committee, who over the past decade have repeatedly adjusted the ODA rules to inflate their numbers. TOSSD’s more balanced governance system should increase its integrity and credibility.

But designing a robust statistical system is hard work. It requires specialist expertise, and professional objectivity. While governance features such as broad representation, political balance and a spirit of compromise may be desirable in the abstract, they risk diluting the specialised professional competence and deep knowledge of the subject matter needed to develop a coherent, robust and workable statistical system.

Yet unfortunately this dilution appears to be what has happened with TOSSD, which displays at least the following four inter-related problems from the point of view of statistical rigour:

1. Lack of provision for reporting of totals

TOSSD collects data only via an activity-level form. There are no data tables requesting total figures on any class of TOSSD expenditure. Published TOSSD aggregates are merely the machine-generated sum of the individual activities reported.

This makes the TOSSD database a collection of examples, without any means of determining their completeness.

Yet statistics are totals: they represent the sum of all the things in defined categories.

You cannot make reliable statistics out of a mere collection of examples. The most basic function of a statistician is to count, which means to establish the correct total of a given thing. This requires work to ensure that all the things within a given category are located and included in the count. But this function is ignored in the TOSSD collection procedure. Its set-up lays no obligation on reporting organisations to ensure that they find and count all relevant activities and certify total numbers for each category of TOSSD expenditure.

This flaw makes it impossible for TOSSD to produce reliable statistics. And in fact, I discovered last year that current TOSSD reporting misses more than 90% of eligible expenditures that are not already included in standard ODA and “other official flows” (OOF) reporting to the OECD. TOSSD states that, for 2022, it captured $81 billion of activities in addition to ODA and OOF. The true total would be well over $1 trillion, since TOSSD covers “big ticket” items that include:

• climate mitigation action in developed countries,

• government-funded basic research “with potential transnational applicability”,

• domestic environmental protection “with spillover effects”,

• domestic pandemic preparedness,

• and all spending on the integration of legal and illegal migrants into “the economy and culture of the provider country such as language training, vocational training, social protection schemes, employment programmes and awareness on national culture” for the first five years of the migrants’ stay.

Unfortunately it is impossible to precisely quantify the shortfall in each case since the TOSSD reporting form does not ask for activities to be identified according to the above categories. The nearest it comes is to identify them using the DAC codes that indicate the sector of the recipient country’s economy being assisted. However, these codes are not designed for regular domestic expenditures in donor countries. The result is that, for example, climate mitigation in donor countries is scored as aid to “Energy” – as if it were support to the energy sector in developing countries; and aid to all immigrants is squeezed into the ODA item of “Refugees in donor countries”.

Using this makeshift categorisation in the third chart here, TOSSD reports that donors spent a grand total of only $16.8 billion in 2022 for support to the “International Public Good” of “Energy” and only $4 billion for that of “General environmental protection”. The true amounts of donors’ domestic spending on these two items, consistent with TOSSD coverage, would be larger by orders of magnitude than those figures. The $36 billion reported for “Refugees in donor countries” at first sight seems closer to the mark, but $31 billion of this was already reported as ODA, and ODA coverage is far narrower than TOSSD coverage. ODA only scores spending for refugees and asylum seekers (not for all immigrants); it only covers the first 12 months of their stay (not for 5 years); and it excludes spending designed to promote the migrants’ integration into the economy and society of the donor country. Obviously, an extra $5 billion does not go anywhere near accounting for all non-ODA spending on assistance to all recent immigrants in all provider countries. Given the breadth and scale of this assistance, the true figure could be in the hundreds of billions.

It is therefore clear that in its current form, TOSSD is only collecting highly incomplete information on individual activities. It offers only examples, not totals; only data, not statistics. Treated as statistics – i.e. as totals of spending in its various categories – its numbers are wildly short of the true amounts. But why is this so?

2. Inherent collection and calculation difficulties

Any new statistical collection poses challenges. And the challenges are greater when the collection aims to capture broad ranges of expenditure using novel definitions, as is the case with TOSSD.

Consider, for example, the qualifications in bold within the following TOSSD items:

• Support to refugees/protected persons/migrants for their integration in the economy of the provider country.

• R&D in areas/issues of global reach e.g. energy, environment.

• Environmental protection activities in the provider country or non-TOSSD recipient countries with spillover effects.

These bolded qualifications are unlikely to appear in donors’ own information systems, and accounting for them would require many arbitrary judgment calls. Even then, reporters’ work would probably not be over, since TOSSD’s list of Pillar II items is repeatedly said to be “non-exhaustive” – leaving them vague and subject to future extension.

There is also a more general problem, which is that TOSSD collects activities whereas many of the items it aims to add to standard ODA and OOF reporting are parts of programmes. Support to migrants would in many cases come not from dedicated funds, but from general programmes available to the general population, such as subsidised training and education, unemployment and sickness benefits, child allowances, housing subsidies, and government grants, loans and financial incentives for various other purposes. Research and development is carried out in all sorts of fields, by many different departments. And environmental protection “with spillover effects” could relate to activities in energy generation, transport, air and water quality, agriculture, forestry, waste disposal and many other fields. In these circumstances it will be hard to convince budgetary authorities to re-order their heads of expenditure and accounting systems to facilitate identification of TOSSD items – especially since TOSSD’s coverage is not required for any other purpose. So even over the medium term it is unlikely that TOSSD coverage will make enough progress to produce useful statistics on anything resembling its current target expenditures.

Further difficulties arise from the fact that TOSSD only collects expenditures. In some TOSSD target areas, much official effort takes the form of revenue reduction rather than expenditure. For example, governments may offer tax deductions for research expenditures, or they may impose laws, regulations, mandates and other measures to achieve environmental goals that both impose costs directly on government agencies and restrict government revenue by imposing costs on private actors, thereby reducing the tax base. These actions demand taxpayer effort just as much as direct expenditure, but would not be covered under current TOSSD rules. This is part of broader problems with TOSSD’s scope that stem ultimately from vagueness and ambiguities in its fundamental concept.

3. Conceptual and definitional issues

TOSSD’s Reporting Instructions describe its four components as follows:

• Pillar I: Cross-border Resource Flows in Support of Sustainable Development

• Pillar II.A: Regional and Global Expenditures in Support of Sustainable Development: Issues specific to developing countries or their populations

• Pillar II.B: Regional and Global Expenditures in Support of Sustainable Development: Issues of global nature

• “Mobilised Private Finance”.

These categories are reasonably distinct, but the problem is that they do not, taken together, form a coherent statistical concept. While they all contribute to sustainable development, they do not represent the total financial flow for sustainable development, since they leave out all private international flows that contribute to sustainable development but which are not mobilised by official action. They are not total official flows for sustainable development either, as they leave out spending by governments of both developed and developing countries on their own domestic sustainable development. Not all TOSSD flows are even aimed at the development of developing countries, since “Pillar II.B covers activities that address issues of a global nature or that provide benefits at the global level e.g. climate change mitigation, general environmental protection.” [Footnote: The credibility of including such “global” expenditures in TOSSD has understandably come into question, for example here, where Mr Serge Tomasi asks “Can we seriously believe that a subsidy paid by the French government to its citizens to encourage them to buy electric vehicles or install solar panels benefits the inhabitants of LDCs such as Senegal or Burkina Faso?”].

This patchwork arrangement is in contrast to traditional DAC reporting where, leaving aside some minor “grey areas”, the categories have clearly defined essential characteristics allowing them to be meaningfully combined. ODA is concessional resource flows for development. OOF is non-concessional flows for development. So ODA plus OOF equals total official flows for development. Private flows are similarly divided into concessional and non-concessional flows for development, so that totals can be formed for total private flows (concessional and non-concessional), total concessional flows (official and private) and total non-concessional flows (official and private). All four sub-categories added together form total resource flows for development.

TOSSD’s disparate categories and lack of a central concept make it hard to define. Its Reporting Instructions offer the following:

1.1 DEFINITION OF TOSSD

The Total Official Support for Sustainable Development (TOSSD) statistical measure

includes all officially-supported resources to promote sustainable development of developing

countries. This includes i) cross-border flows to developing countries and ii) resources to support development enablers and/or address global challenges at regional or global levels.

Various terms included in this statement are then expanded upon. However, saying what something includes is not a definition; it is only a statement of coverage, and an open-ended one at that. And the explanations of terms themselves require explanation. It emerges that “officially-supported” is being used to cover “official”, though they are different [“Officially supported” flows are private flows backed by finance, guarantees or insurance from the official sector]; and that domestic expenditure by developing country governments is excluded, though the definition does not say so. The explanation of “resources to support development enablers and/or address global challenges at regional or global levels” is extremely vague, yet clearly brings in resources beyond those intended mainly “to promote the sustainable development of developing countries”.

TOSSD’s difficulties in defining itself may be partly the result of its long gestation period, and its understandable efforts to meet the expressed coverage needs of different groups. Whatever the cause, TOSSD remains vague as a statistical concept. While it may be possible to go back to basics and clearly specify such a concept, this would almost certainly then require major changes to its definition and coverage.

4. Valuation issues

Meanwhile, given its current scope, there is a valuation problem within TOSSD’s measurement. In principle, it aims to score dollars that promote “the sustainable development of developing countries”. But looked at through that lens, different parts of TOSSD have different dollar values. Some elements of TOSSD would have to be discounted to arrive at their value as sustainable development finance, and only the discounted figure could be included in a total claiming to represent “resource flows to promote the sustainable development of developing countries”.

The discounting would have to be applied to most of the contents of Pillar II.B, which covers “Issues of a global nature”, since the benefits of this finance are shared globally, and may even accrue mainly to the provider country rather than to developing countries.

Discounting would also be needed for some of the contents of Pillar II.A (“Issues specific to developing countries or their populations”) such as:

• “Support to refugees/protected persons/migrants for their integration in the economy of the provider country”. Such support benefits the provider country, and the migrant, but disadvantages the migrant’s country of origin by making it less likely that he or she will

return there, and thus permanently depriving that country of his or her skills. From the point of view of dollars for the sustainable development of source countries, this would require a negative adjustment.

• “Imputed (indirect) student costs for tuition in provider countries provided to nationals of TOSSD recipients”. These subsidies of education costs in developed countries primarily benefit the provider country if the student remains there to work after his or her studies, as often happens, so the full amount should therefore not be scored as promoting the sustainable development of the students’ countries of origin.

TOSSD principals appear to be aware of the potential for reported TOSSD expenditures to be worth less than their face value in dollars in terms of promoting sustainable development. They have tried to address this issue by insisting in their Reporting Instructions that all expenditures reported to TOSSD must meet a wide variety of sustainability and other criteria:

In line with inherent thrust of the SDGs – to promote a more sustainable, equitable and prosperous world for all people – this statistical framework assumes that all resources captured therein should be provided consistent with prevailing global and regional economic, environmental and social standards and disciplines, with development co-operation effectiveness principles, as well as with the United Nations Charter and International Law. These safeguards ensure that TOSSD-eligible investments are sustainable, promote equal opportunities and rights, guard against negative environmental, social and climate impacts and risks, and – where necessary – limit damage through mitigation measures. Particular attention is paid to commercially motivated resource flows – such as officially-supported export credits, subsidies and resources mobilised from the private sector – and on their compliance with global standards and disciplines, including to ensure that they do not create trade distortions. It is also assumed that providers of TOSSD seek to ensure that the financing of global and regional expenditures does not have a crowding-out effect on their country-specific development co-operation to TOSSD recipients. When granting scholarships or hosting students from TOSSD-recipients in their education and training institutions, provider countries should take into consideration whether the partner country has put in place incentives to minimise brain drain in developing countries.

The logic here is that any expenditure that met all these criteria could be scored in TOSSD at 100% of its value. However, no guidance is offered on how to discount flows that only partially meet the criteria. And in any case, it is not the role of a statistical system, or of statistical reporters, to enforce such standards, and it would be impossible in practice to do so on a consistent basis. TOSSD’s valuation problem remains, and needs to be addressed within a fundamental rethinking of its concept and coverage.

A possible path forward?

Solving TOSSD’s problems is a matter for its International Forum, but one promising recent idea [Footnote: 3 The idea is due to Mr Tomasi, whose work has already been mentioned. See here for details.] is to simplify it by dropping the distinction between cross-border and non-cross border flows and having only two categories: development finance, and contributions to global public goods. This would help remedy the complexity and vague borderlines of the current TOSSD categories. Admittedly, developing a credible and broadly accepted GPG measure would require the involvement of high-level experts on this topic, and fundamental conceptual work. Also, as a practical matter, it would likely change TOSSD’s existing coverage of GPGs, so that the GPG data it has collected to date would

be useless under the new definition. However, in practice this would not be a serious disadvantage, since as we have seen, those data are useless even on the current definition.

Taking this idea a little further, it may be useful to split off the GPG work and create a separate body to work on it. This would have several advantages. First, it would allow the new body to dispense with the difficulties created by TOSSD’s “jamming” GPGs into a measure of development support supposedly focused on the recipient perspective. Why should developing countries care about the amounts that donors spend on basic research, or on integrating migrants (except in the negative sense that this promotes “brain drain”)?

A separate body would also make it easier to attract the specialised expertise necessary to formulate a concept, definition and scope for a GPG statistical collection that could command general acceptance. Much work has already been done in this field, and there is a large academic and institutional literature on it. Developing a credible measure is really a job for a small group of experts in the field, and they would need to be free to decide whether, and in what way, the measure should be linked to development finance.

A further advantage of this approach is that it would allow TOSSD to concentrate on the important task of improving data on flows to developing countries. It has already launched a “data review mechanism” to work on this. But it is a huge job to conciliate the donor data on aid and other resource flows with recipient records. Much donor funding flows through intermediaries, who may or may not be keeping recipient countries fully informed of their activities. There may be delays between donor expenditures and beneficiary receipts or practical actions on the ground. Recipients’ records on development activities may only contain their own expenditures, and be organised on a different basis from provider country information. Genuine reconciliation of provider and recipient perspectives will require a great deal of exacting work, and will require TOSSD participants’ full attention.

The results, however, could represent a major breakthrough in the transparency of information on international resource transfers. For decades, researchers have puzzled, and debated, such questions as “how much aid really reaches beneficiary populations?” “how much aid disappears in administrative overhead, or flows to developed country firms, consultants or experts?” “how much aid is under the control, or even within the knowledge, of developing country governments?” and “how much money and resources do developing countries have to contribute themselves to make projects work?”. Serious data have always been lacking to answer such questions, and TOSSD might finally be able to provide such data.

Splitting off the GPG work from the TOSSD process would also allow both undertakings to pay attention to statistical quality criteria such as accuracy, reliability, clarity, completeness, and consistency over time and across reporters. These features are vital to the success of any statistical system, but have been neglected so far as TOSSD has been under perpetual development in an effort to be “all things to all men”.

Conclusion

TOSSD has aimed high, but fallen short. It is currently missing over 90% of the expenditures it aims to add to standard OECD data on official developmental flows. It has no realistic chance of ever filling this gap because most of the flows it is trying to add are not discrete activities but integral parts of broad government programmes. Moreover, its exact coverage is not clear because TOSSD’s basic concept remains woolly, and its specified components do not form a coherent whole. TOSSD purports to measure financing for the sustainable development of developing countries, but several of its current items would require discounting to arrive at their true dollar value as such financing. These problems mainly affect its attempt to measure global public goods. A possible way forward would be to create a separate structure to develop a measure of GHG financing, so that TOSSD could focus clearly on resources dedicated to developing countries.

Comment by Hedwig Riegler:

I. General remarks

Deterioration of the organisational and presentational culture of development finance statistics

When I entered the OECD/DAC Statistics “world” in the mid-1990s, I was awed by the impression that I was dealing with honest efforts to depict a quantitative reality as closely as possible, and I felt that this was a place where I belonged. That world is now long gone. Nowadays, when I look at the “facts and figures” presented in this field, it frequently strikes me that they are no longer trying to capture a reality but only to sell a product – a political narrative in statistical disguise.

This seems to me part of a general deterioration of the culture of public administration. We seem to be living in times of a great transformation, from truly and professionally trying to implement policies to merely pretending to do so, while putting the real effort into marketing the product instead of improving it. Thus the handicraft of doing statistics is being replaced by the arts of selling them. This trend is particularly pernicious in the field of development finance statistics, where donors wish to appear to be meeting high-profile aid and resource flow targets, yet are unwilling to spend real money to do so.

TOSSD is no exception, but rather a perfect example of this transformation from producing facts to producing political narratives. The TOSSD collection system reflects precisely this: it rests on political campaigning strategies instead of the professional principles laid down for statistics – especially the principle of maintaining independence from political interference. And while ODA rule-making has in recent years increasingly been in the hands of political decision-makers, this has been the case for TOSSD rule-making from the start. Therefore, in both cases, but especially with TOSSD, I doubt the sustainability of products that have been created without applying proper statistical craftsmanship.

II. Specific comments on Simon Scott’s text above

“But these measures are little known and seldom used …”:

The measures of Official Development Finance and Total Resource Flows for development are, in fact, highly relevant and useful. But since – as opposed to ODA – no firm quantitative targets have been attached to them for many years, all attention has been focused on ODA, as if it were the only relevant finance for developing countries.

Non-ODA flows have been captured by DAC Statistics ever since their beginnings in the 1960s, but further system development and modernisation was neglected by the DAC for too long. Actually, the original idea of TOSSD (originally TOSD – Total Official Support for Development) was to remedy this by reviving and modernising the capture of non-ODA finance and incorporating new and innovative financing – an excellent idea. But then this undertaking derailed for various reasons, in my opinion mainly because it was understood as a political endeavour and negotiation process, and not as the building of a modernised metric that would merit the name of “statistics”. Part of the lopsided set-up in my understanding was a felt need to jump on the bandwagon of SDG monitoring and to win acceptance, if not adoption, by the UN system, especially the UN Statistical Commission – an endeavour that failed.

“Lack of provision for reporting of totals”

In my view, the key issue here is NOT the lack of provision for totals reporting, but rather the lack of a clear definition of totals and sub-totals. In fact, I consider this as the “loadbearing structure” of any statistical system: a precise definition of the overall total (in this case, all items belonging under the heading and conforming to the essential idea of TOSSD) and all sub-totals into which the overall total can be divided. Mathematically speaking, we would use the term “set”, and the theory of sets actually applies: the universal/overall set would be divisible into relevant sub-sets and sub-sub-sets.

Also, the issue for me is not so much whose machines do the adding up, but the absence in TOSSD of a precise aggregation prescription understood in the same way by both the reporting entities and the TOSSD Secretariat in its data-collating role. Whether or not the Secretariat does aggregations, a mechanism must be in place for reporting entities to do the same aggregation for their own reported dataset, to compare and verify results before overall data presentations are published. Aggregation rules must be crystal-clear in reporting instructions. And in a solid statistical set-up they would be unmistakably articulated, starting from the definition of the overall “target set” and proceeding by unambiguous indications to the sub-sets and sub-sub-sets into which the total could be broken down.

In TOSSD, this is left ambiguous: e.g. several instructions and headline graphics suggest that “officially mobilised private finance” is a subset of TOSSD. But in table presentations, the figure for this finance is stated at the end as a separate category. It remains unclear whether it is included or not in “total TOSSD”. In statistics, there should be no ambiguity: an item is included or excluded in a set – there is no in-between, because statistical collections are “closed sets”.

Contrary to this fundamental statistical principle, TOSSD is still an “open set”, which means items are added continuously. This results in meaningless numbers and, especially, meaningless time series, since increases in such a system will be primarily due to an expansion of the collection and not to any increase of the resources made available for the given purpose.

Of course, from time to time, revisions and updates are implemented in “closed” sets of robust statistics. When necessary, such collections are “opened up”, revisions are made, and if they imply significant changes, past data are revised retroactively in order to maintain the comparability and consistency of time series. These are normal procedures in real statistical systems. With TOSSD, by contrast, there has been constant tinkering with coverage and examples. TOSSD annual figures thus reflect the state of the rules at the time, the state of the political negotiation process, rather than any actual quantitative evolution of sustainable development finance over time. TOSSD will be incapable of delivering the latter until it adopts professional statistical standards in developing and revising methodology and data.

Even in a “closed” set, with stable coverage, it may still be difficult to capture 100% of the items that belong, but in solid statistics, metadata on the coverage will estimate how close to 100% the collection has come, and for reliable statistics coverage must be a high percentage. A coverage of way below 50%, as Simon Scott suggests for TOSSD, cannot justify calling output numbers “statistical figures”.

“Lack of a unifying statistical concept”

In my view this relates directly to what I have just said about the lack of a clear concept of the subsets involved in the overall set “TOSSD” and their relation to each other. In a robust collection system, the headline figure for TOSSD would be the top (umbrella) category, under which all sub- categories could be legitimately subsumed and where each tier of sub-categories is shown in its correct relation to the others, as in an organigram or maybe a family tree.

However, the TOSSD collection is a hotch-potch of different things whose relation to one another is unclear, and it also fails to give the “complete picture”, despite its claims to the contrary, since it not only fails to reach acceptable coverage on the categories included, but also deliberately leaves out the vast majority of the highly relevant private financing for development. So, what is TOSSD aiming to show actually? What exact quantitative question does it answer? Because that is what statistics are for: answering concrete quantitative questions. The original idea of TOSSD could have been worth following, but given its politically-steered evolution it has not matured into a “statistic”.

In my view there is little sense in going into all the detailed technical deficiencies of TOSSD, because the absence of a solid foundation and “loadbearing structure” makes these inevitable. No matter how much is invested in the decoration of a house built on shaky ground, it will be doomed to collapse. And although TOSSD now bids fair to turn into a “skyscraper”, it still has no solid foundation. As explained in my previous article, TOSSD aspires to measure a much-too-broad, fuzzy concept, which is left undefined in too many aspects to become a true statistical quantity. And the motivation probably never has been to really build a statistic, but rather to establish a political negotiation process that would justify and “validate” political narratives by underpinning them with numbers that look like statistics.

“… it is not the role of a statistical system, or of statistical reporters, to enforce such standards, and it would be impossible for either of them to do so.”

Here, I couldn’t agree more. In particular, when the “metric” is to measure “finance”, which in this case means dollars committed or spent, the value-added it can supply is just further information – such as the intention with which finance was allocated, the nature of goods or services supplied, financial terms, tying etc. By contrast, statistical reporters, and a statistical system including its administrative (secretariat) structure, can never deliver a screening of each financed activity against quality standards. Quality requirements can never be met effectively in practice by statistical procedures. They are unmanageable for any of the actors in a statistical system, and quality assurance of the activities captured by finance statistics is not the role of those statistics but of other functions and instruments such as appraisals, approval criteria, and monitoring and evaluation systems. In reality, the “quality standards” injected into TOSSD are mere political persiflage designed to inflate the significance of its work, while neglecting its basic professional requirements, such as the design of a robust collection structure, and specification of clear and consistent definitions.

Can TOSSD’s problems be solved?

The idea of “having only two categories: development finance, and contributions to global public goods” picked up by Simon Scott from Mr Tomasi’s suggestion is a promising new approach, but in my opinion will only work if the global public goods category is handled in a separate system by a separate forum. Its nature is sufficiently different from the other financial flows in TOSSD that it has to be approached with different procedures and instruments. Moreover, financial contributions to global public goods are embedded in many different government programmes, and there is at present no clear-cut financial structure in official budgets that would allow for a manageable derivation of reportable amounts or identification of eligible portions. For these reasons, trying to combine development finance and contributions to global public goods in one system will lead, as we can already see, to both categories suffering a serious loss of precision and statistical quality.

My firm conviction is that these problems can only be solved by starting from scratch: there is no way of patching up the current TOSSD system so as to turn it into a truly statistical measure. The present architecture of TOSSD is not that of a “sound building”, and adding to or replacing its existing elements will just further destabilise its rickety construction. And, honestly, I would not want to invest time in such an undertaking.