448 reporting organizations in Kenya. Four of them account for half the money. And the first version of this analysis got the headline wrong.
The development policy conversation worries about aid fragmentation — too many donors, too much coordination overhead. Weigh the donors by spending instead of counting them and the picture changes: a handful of funders, led by the multilateral banks, control most of the money. It is a concentration story. It is just not the concentration story first reported here.
Count organizations reporting IATI activities: you see fragmentation — 253 to 448 per country. Weight by spending, with each activity credited to a country only in the proportion its publisher declares: you see concentration — four organizations reach half the money, 12 to 15 reach 80%.
| Country | Orgs | 50% in | 80% in | Top 2 share | Gini | Orgs <0.1% |
|---|---|---|---|---|---|---|
| Kenya | 448 | 4 | 15 | 41.6% | 0.950 | 88% |
| Uganda | 427 | 4 | 14 | 35.5% | 0.949 | 89% |
| Ethiopia | 358 | 4 | 15 | 38.0% | 0.945 | 87% |
| Tanzania | 290 | 4 | 12 | 33.5% | 0.942 | 87% |
| Rwanda | 253 | 4 | 12 | 32.3% | 0.932 | 81% |
IATI activities can be tagged to many recipient countries, each with a percentage. A naive country filter returns every tagged activity at full value. Weighting by the declared percentage removes 69–79% of the apparent spend in these five countries — $1,156 billion becomes $274 billion — and it removes it unevenly. Publishers of large global programs lose most; publishers of country projects lose nothing.
| Country | Top 2, unweighted | Top 2, weighted | #1 unweighted | #1 weighted | Spend belonging elsewhere |
|---|---|---|---|---|---|
| Kenya | 66.1% | 41.6% | Dept. of State (44%) | World Bank (26%) | 79% |
| Uganda | 64.7% | 35.5% | Dept. of State (43%) | AidData* (23%) | 79% |
| Ethiopia | 59.4% | 38.0% | Dept. of State (36%) | World Bank (23%) | 69% |
| Tanzania | 66.5% | 33.5% | Dept. of State (43%) | World Bank (22%) | 78% |
| Rwanda | 63.6% | 32.3% | USAID (48%) | World Bank (22%) | 79% |
The U.S. Department of State falls from first place in four countries to outside the top six in all five. The World Bank becomes the largest reported funder in four of five. USAID stays second or third, on a much smaller base. The corrected landscape is multilateral-led with a strong U.S. bilateral presence, not U.S.-dominated.
Eight organizations appear in every country’s weighted top 15. The aid landscape looks diverse when you count organizations, but the dominant funders are the same across the region.
| Organization | Weighted spend | Countries | Note |
|---|---|---|---|
| World Bank | $56.8B | 5/5 | Largest in KE, ET, TZ, RW |
| USAID | $37.8B | 5/5 | |
| AidData | $27.8B | 5/5 | Research org; reports third-party data. Largest in UG |
| African Dev. Bank | $18.8B | 5/5 | |
| UK (FCDO) | $14.7B | 5/5 | |
| Global Fund | $13.7B | 5/5 | |
| Sweden (Sida) | $7.5B | 5/5 | |
| UNICEF | $6.8B | 5/5 | |
| HHS (CDC, NIH) | $6.7B | 4/5 | Not in Ethiopia top 15 |
| Germany (BMZ) | $5.8B | 5/5 | |
| UNHCR | $4.6B | 4/5 | |
| Gavi | $3.8B | 3/5 | |
| Denmark (MFA) | $3.7B | 3/5 | |
| U.S. Dept. of State | — | 0/5 in top 6 | Was #1 in four countries before weighting |
AidData, a research lab at William & Mary, appears among the top five funders in every country and is the largest single “donor” in Uganda. It is not a donor — it publishes historical aid data into IATI from funders (especially China) that don’t publish themselves. Its entries represent real flows, reported by a third party, and may overlap with activities reported by the original funder.
Weighting does not remove this artifact, because AidData’s activities are mostly single-country. It is a different measurement-layer problem: the identity of the reporter is being read as the identity of the funder.
The uncorrected version passed a full analysis, a written brief, three charts and publication. The artifact was found only when the same tool was pointed at Pacific island states, where a $90 billion “donor” to Tonga (national GDP under $1 billion) was impossible to miss. In East Africa, where U.S. spending genuinely is large, the same artifact produced numbers that looked plausible.
Nothing in the output flags the omission. A country filter that returns “$262 billion of aid to Kenya” is not wrong in any way the database can detect. Only a reader with an external sense of scale can catch it — and for East Africa, that sense of scale was not enough. This is the measurement layer operating on the analyst. See the companion analysis for the Pacific: Who Funds the Pacific?
Data from d-portal.org, activity and country tables joined on activity identifier. ~97,000 IATI activities with positive spend across five East African countries (Kenya, Uganda, Rwanda, Tanzania, Ethiopia) from ~1,700 reporting organizations. Weighted spend = activity lifetime spend × declared country_percent / 100; single-country activities carry 100%. 19–29% of activities per country are multi-country; they account for 69–79% of unweighted spend. Spend values are lifetime totals, not annual. Organizational identity uses reporting_ref — PEPFAR spending is attributed to the Department of State (the IATI publisher). Unweighted figures are retained for comparison. Tool: donor_concentration_weighted.py.