Work on the third DFI Transparency Index is underway, in preparation for launch in summer 2027. The Index ranks the transparency of the world’s leading development finance institutions (DFIs) and encourages improvement. The 2025 DFI Transparency Index found some improvements in transparency across both sovereign and non-sovereign portfolios since 2023. However, serious gaps remained, especially around disclosure of development impact, private capital mobilisation, and accountability to affected communities.

In 2026 we conducted a methodology review to improve the rigour of the DFI Transparency Index, while being careful to maintain a degree of comparability between the 2023, 2025 and 2027 editions. We updated the methodology following an in-depth internal review of the Index, in addition to consideration of written feedback from stakeholders.

Methodology updates

We have set out the key changes below. Full details can be seen in the methodology paper.

1. Clarification of the completeness penalty system

Comprehensive disclosure of investment activity is essential for stakeholders to understand the full scope of DFI operations. The previous completeness test functioned primarily as a proxy for timeliness and could fail to identify institutions that consistently disclose only a portion of their investments.

For the 2027 Index, the sample will use projects approved in 2026, but the completeness test will use 2025 data, as annual reports for the sample year will not be consistently available when the analysis is conducted. DFIs are expected to disclose at least 50% of the activity reported in their annual report. Where disclosure falls below this, we will contact the institution to clarify the discrepancy during the engagement period. If no satisfactory explanation is provided, a 50% penalty to project-level scores will be applied. A 100% penalty will be applied where no projects from the relevant year are disclosed at all. We will apply the test proportionately in its first implementation and refine it over future cycles. See section 6 in the methodology paper for more detail.

2. Risk segmentation of environmental and social disclosure analysis

Previously, we assessed environmental and social (E&S) document disclosure using a general sample of projects with no intentional sampling by risk level. As low and medium risk projects are more common, our sample typically captured mainly medium risk projects, meaning a DFI disclosing well for high risk projects but not medium risk ones could perform similarly to one disclosing for neither.

We now segment our analysis into two risk-aligned samples: one for high risk projects (Category A direct investments and Category FI1 investments), and one for low and medium risk projects. The affected indicators (Indicator 31 and Indicator 32) are analysed once per sample, with indicator weights distributed evenly between the two. This change gives a more accurate picture of DFI disclosure practices across different risk levels.

3. Change to IATI scoring protocols

Previously, a DFI publishing relevant information solely to the International Aid Transparency Initiative (IATI) received no credit for doing so, as points were only awarded where IATI publication was consistent with another format. We are removing this requirement so that IATI publication is assessed independently. DFIs can now score points for publishing to IATI even where the same information is not available elsewhere.

4. Reducing the first round project sample size

The first round of assessment is indicative and does not contribute to final scores. Given our experience shows a high degree of consistency in disclosure practices across an institution’s portfolio, we are reducing the first-round sample size from 15 to 10 projects. This reduces the burden on both the assessment team and DFIs without compromising the aims of the round.

5. Introducing AI-assisted evidence search and screening

We are introducing a limited and clearly defined use of AI tools to support document identification and screening. AI will be used only to assist the initial identification and screening of sources and not to score indicators, make evaluative judgements, or override human judgement. All evidence identified by AI tools is subject to mandatory team review, and only verified evidence is used.

Timeline for the 2027 DFI Transparency Index

Work on the 2027 Index is scheduled as follows:

October 2026 Database Building – First Round

We will extract data from DFI investment lists and databases to identify investments from 1^st January 2026 to 30^th September 2026. We will select our samples at random from this list.

November to December 2026 Data Collection and Analysis – First Round

We will assess DFIs’ organisation and project-level disclosure from the first sample of ten projects. This round of analysis will not contribute to the final score of DFIs in the index; it is intended to give DFIs an indication of their performance and to identify areas for improvement.

January to February 2027 DFI Review

We will send first round scores directly to DFIs in January to give them an opportunity to understand our assessment of their performance and provide feedback on our findings. This will also provide time for DFIs to make improvements to their disclosure in line with our initial assessment. DFI feedback will be considered prior to the second round of data collection and analysis.

February 2027 Database Building – Second Round

We will extract data from DFI investment lists and databases to identify investments from 1^st January 2026 to 31^st December 2026. We will select our samples at random from this list.

March to April 2027 Data Collection and Analysis – Second Round

We will assess DFIs’ organisation and project level disclosure based on the second sample of fifteen projects. This round of analysis will be the only round that contributes to index scores.

May to June 2027 Report Production

We will rank the DFIs included in the Index according to their performance, and present results in a public report alongside detailed analysis of the assessments.

June/July 2027 Launch of report

We will launch the Index report at a public event in summer 2027.

DFIs to be assessed

As part of this review, the total asset size thresholds for inclusion have been raised. The methodology paper sets out the full selection criteria.

The DFIs to be assessed in the 2027 Index are set out below. This is the same cohort of DFIs that were assessed in 2025.

Multilateral – Sovereign

• African Development Bank (AfDB)
• Asian Development Bank (AsDB)
• Asian Infrastructure Investment Bank (AIIB)
• Development Bank of Latin America and the Caribbean (CAF)
• European Bank for Reconstruction and Development (EBRD)
• European Investment Bank (EIB)
• Inter-American Development Bank (IDB)
• Islamic Development Bank (IsDB)
• New Development Bank (NDB)
• World Bank (WB)

Multilateral – Non-Sovereign

• African Development Bank (AfDB)
• Asian Development Bank (AsDB)
• Asian Infrastructure Investment Bank (AIIB)
• Development Bank of Latin America and the Caribbean (CAF)
• European Bank for Reconstruction and Development (EBRD)
• European Investment Bank (EIB)
• IDB Invest
• International Finance Corporation (IFC)
• Islamic Corporation for the Development of the Private Sector (ICD)
• New Development Bank (NDB)

Bilateral

• Austrian Development Bank (OeEB) (Austria)
• Belgium Investment Company for Developing Countries (BIO) (Belgium)
• British International Investment (BII) (United Kingdom)
• Finnfund (Finland)
• German Investment and Development Company (DEG) (Germany)
• Dutch Entrepreneurial Development Bank (FMO) (Netherlands)
• Impact Fund Denmark (IFD) (Denmark)
• Norwegian Investment Fund for Developing Countries (Norfund) (Norway)
• Promotion and Participation Company for Economic Cooperation (Proparco) (France)
• Swedfund International AB (Swedfund) (Sweden)
• Swiss Investment Fund for Emerging Markets (SIFEM) (Switzerland)
• US International Development Finance Corporation (DFC) (United States)