How the numbers are calculated¶
Every figure in the portal traces back to the Global Database maintained by the GPEDC Joint Support Team. Most are published there and shown unchanged. A few are worked out by the portal, and this page says which, so you can tell the difference.
Read it when a number looks different from one you have seen elsewhere, or when you want to know exactly what an average covers.
What is published and what is calculated¶
| Figure | Where it comes from |
|---|---|
| An individual country's or partner's indicator score | Published in the Global Database, shown unchanged |
| Group averages by income group, region or organisation type | Published in the Global Database |
| Untied aid | Compiled by the OECD from its Creditor Reporting System |
| Dimension averages | Calculated by the portal — see below |
| The direction and size of a gap ("+9% above the group average") | Calculated by the portal from the two published figures |
Where the portal calculates something, it says so on screen.
Group averages¶
Group averages are not a simple mean of the countries or partners in the group. For measures based on money — how much funding is recorded on budget, how much is disbursed on schedule, how much uses country systems — the published average is weighted by financial volume. A partner disbursing $2 billion counts for more than one disbursing $2 million, because the question is what happened to the money, not what the typical organisation did.
For measures not based on money, the average treats each country or partner equally.
This is why a group average can sit some way from the middle of the individual scores, and why the figures here match published GPEDC tables.
Two measures are the exception, and the portal calculates their group figures itself because the Global Database does not carry them:
- Untied aid — total untied ODA over total ODA commitments across the DAC members that took part in the round.
- Quality of PFM systems, which is derived from PEFA assessments rather than the questionnaire.
Dimension averages¶
The four dimension scores — the ones in the coloured cards at the top of a profile, and on the Compare radar — are the portal's own summary. They are not part of the official methodology, and the portal labels them as such wherever they appear.
Each is the average of the headline indicators in that dimension. Where a component has a composite indicator, the composite is used and its parts are left out, so a component with four sub-measures does not outweigh one with a single measure.
Three kinds of measure are excluded from dimension averages:
- Direction-of-travel measures, such as the PFM trend. Improvement, no change and decline are not a scale of 0 to 1, and averaging them into one would distort the result.
- Categorical ratings, such as the transparency grades, which the methodology states are not directly comparable.
- Missing values, which are skipped rather than counted as zero.
Comparison groups¶
Partner countries are compared with others in the same World Bank income group. This replaced comparison against a global average, which set very different economies against one another.
Development partners are compared with one of three groups:
| Group | Who is in it |
|---|---|
| DAC members | Bilateral providers who are members of the OECD Development Assistance Committee |
| Multilateral partners | Multilateral organisations, including the development banks |
| United Nations entities | UN funds, programmes and agencies |
A partner is placed in exactly one group, and the same assignment is used everywhere — the map, the profiles, and the downloadable PDF profiles all compare against the same set.
Partners reporting in fewer than five partner countries are shown without a group comparison. A score from one or two countries is not comparable with one drawn from thirty, and presenting them side by side would read as a judgement on practice when it is an artefact of how little was reported.
When two figures for the same thing differ¶
A handful of measures appear in more than one place with different values. This is intended, and each is labelled:
- Untied aid. The headline figure is the partner's published untied share across all recipients of its development co-operation, for the OECD's latest year (2024). The map shows the share in each ODA recipient from the CRS analysis compiled for the round, which carries its own reference year (2023, or 2022 for a few recipients). The two are the same statistic for different years, so they do not match, and no average of the countries on the map is meant to reproduce the headline.
- A partner's own score against its group average. Both are published figures, but the group average may be weighted by financial volume while the individual score is not.
- Coverage counts. "Reported in 28 partner countries" counts the countries a partner reported on to the round. "120 ODA recipients where the CRS records a commitment" counts every recipient the OECD records a commitment to, most of which are outside the round. Neither is a subset of the other.
Rounds and years¶
The current round covers 2023–26 and is the reference period for every score. Where a previous-round figure exists, it is shown alongside, and the comparison is against 2018.
Two sets of measures keep their own years, because the institutions that compile them publish on their own cycle:
- Untied aid — 2024, compared against 2017.
- Transparency ratings — 2025, compared against 2018.
These years are shown on the measures themselves so a comparison is never read against the wrong period.
Missing data¶
A missing value is never treated as zero. It is left out of averages, comparisons and lists.
This matters most where a measure has several parts. If a partner reported on three of four elements, the score reflects the three reported; the fourth does not drag it down. Where too little was reported for a result to be meaningful, no result is published, and the portal says why rather than showing a low score.