Sector codes, keywords, and Rio markers identify almost entirely different activities in IATI data
“How much climate finance flows to developing countries?” is one of the most politically consequential questions in international development. The answer depends on something rarely made explicit: how you identify climate finance in the first place.
We applied three methods to the same IATI dataset for Kenya. Sector codes (DAC 5-digit codes for environmental and energy sectors) identified 2,126 activities. Keywords (“climate,” “adaptation,” “resilience,” “solar” etc. in activity titles) identified 1,331 unique activities. Rio markers (the OECD DAC policy markers for climate mitigation and adaptation) tagged only 37% of climate-sector activities.
| Method | Activities | Commitments (USD) |
|---|---|---|
| Sector codes | 2,126 | $44.4B |
| Keywords (unique) | 1,331 | $6.7B (climate only) |
| Rio markers (tagged) | 156 of 419 checked | — |
The most striking finding was not the different totals but the minimal overlap. Of 2,502 unique activities identified by either sector codes or keywords, only 186 appeared in both sets — 7%. The two approaches essentially identify different pools: one captures activities in environmental sectors regardless of title language, the other captures climate-titled activities across all sectors including agriculture, health, and governance.
| Category | Activities | % of union |
|---|---|---|
| Sector codes only | 1,174 | 47% |
| Both methods | 186 | 7% |
| Keywords only | 1,142 | 46% |
The Rio marker approach — the basis for official climate finance reporting to the OECD — has a deeper structural problem. Across five East African countries, 60–69% of climate-sector activities are invisible to Rio markers because reporting organizations don’t fill them in. The gap is systematic: it splits along donor lines.
| Donor | Activities | Marker fill | Mitigation | Adaptation |
|---|---|---|---|---|
| USAID | 297 | 0% | 0 | 0 |
| Sweden | 123 | 63% | 48 | 48 |
| UK/FCDO | 103 | 31% | 23 | 20 |
| Netherlands | 72 | 90% | 48 | 9 |
| Practical Action | 62 | 0% | 0 | 0 |
| AFD (France) | 33 | 0% | 0 | 0 |
| Finland | 28 | 93% | 10 | 6 |
| GEF | 25 | 0% | 0 | 0 |
| Green Climate Fund | 20 | 100% | 19 | 12 |
| Germany (BMZ) | 19 | 100% | 9 | 8 |
USAID — the largest reporter of environmental activities in every East African country examined — has zero marker completion across nearly 300 activities per country. The Global Environment Facility, a major multilateral climate fund, also reports zero markers. Meanwhile, the Green Climate Fund, Germany, and the Netherlands achieve 90–100%.
This means the measurement method effectively determines which donors’ activities count. Rio markers systematically undercount US and multilateral climate finance while fully capturing European bilateral contributions.
Among activities that do carry climate markers, another measurement choice shapes the total: whether to count “significant” activities (where climate is a secondary objective) alongside “principal” ones. In the $100B accounting debate, whether to count significant-objective activities at 100% of their value is one of the most consequential methodological choices.
| Marker | Principal | Significant | Change |
|---|---|---|---|
| Mitigation | 129 | +77 | +60% |
| Adaptation | 41 | +94 | +229% |
Including “significant” alongside “principal” increases mitigation counts by 60% and adaptation counts by 229%. For adaptation — where climate is more often a secondary objective of water, agriculture, or resilience projects — the significance threshold nearly triples the total.
The measurement inconsistency is not specific to Kenya. Across all five East African countries, the same structural pattern holds: different methods, different answers, and 60–69% of activities invisible to Rio markers.
| Country | Sector codes | Keywords | Env checked | % with markers |
|---|---|---|---|---|
| Kenya | 1,660 | 612 | 784 | 35% |
| Uganda | 1,024 | 368 | 617 | 31% |
| Tanzania | 1,302 | 285 | — | 43% |
| Ethiopia | 1,102 | 449 | 616 | 32% |
| Rwanda | 765 | 181 | — | 77%* |
*Rwanda’s higher rate reflects its smaller activity pool, where European marker-using donors represent a larger share.
The climate finance measurement problem in IATI mirrors the fragmentation measurement problem this series has documented: the answer depends more on how you count than on what’s being counted.
For climate finance tracking: Official OECD climate finance figures, which rely on Rio markers, systematically undercount contributions from organizations that don’t fill in markers — notably the largest bilateral donor (US) and several major multilateral funds. The “how much climate finance?” question doesn’t have a single answer in IATI data.
For IATI data users: Sector codes and keyword searches identify almost completely different activity pools (7% overlap). Researchers using one approach will reach fundamentally different conclusions from researchers using another, even with the same underlying data.
For data quality reform: Making Rio markers mandatory in IATI reporting would improve comparability but wouldn’t resolve the underlying definitional question of what counts as climate finance — a road improvement in a flood-prone area is either adaptation or infrastructure depending on who’s coding it.
Data was retrieved from the Code for IATI Datastore API (activity and XML endpoints) in September 2026. Sector-code analysis used DAC 5-digit codes for environmental (410xx), energy (232xx), and conservation (14015) sectors. Keyword analysis searched activity titles for climate-related terms. Policy markers were extracted from XML activity records, with significance levels coded as principal (2), significant (1), not targeted (0), or missing. Donor marker analysis examined activities across five environmental and energy sectors per country. Overlap analysis compared IATI identifiers across sector-code and keyword result sets.