Policy Analysis · MDB Reform Monitor · March 2026
The AfDB’s Self-Evaluation System and the Evidence Behind the Number
In the 2020–2021 validation cycle, AfDB’s Independent Development Evaluation (IDEV) reported that 94% of completed sovereign operations achieved a positive performance rating. This figure has been cited in donor briefings and institutional self-assessments as evidence of strong portfolio performance. It is not. It is evidence of a rating architecture that is structurally incapable of producing a below-satisfactory aggregate result. The cases in this note show what that means in practice: a power plant that will not earn back its cost of capital, rated Satisfactory; a bridge project that ran ten years instead of three, rated Satisfactory; a task manager who wrote that his own sub-rating was “possibly misrepresentative” — and kept it.
Part I — The Architecture That Produces 94%
1.1 What IDEV Actually Is — and Is Not
Before interpreting the 94% figure, the architecture that produces it must be understood. IDEV is not the equivalent of the World Bank’s IEG. The distinction is operational, not cosmetic.
| Dimension | World Bank IEG | AfDB IDEV |
|---|---|---|
| Primary rating function | Generates independent outcome ratings on completed projects — the IEG rating stands as a separate verdict alongside the ICR self-rating | Validates management’s Project Completion Reports (PCRs) — produces a PCREN on the quality of management’s self-assessment |
| Independence of outcome verdict | IEG rates projects independently; divergence is tracked and published | IDEV ‘validates’ the PCR; IDEV’s assessment is of whether management’s rating is plausible — not an independent verdict on project outcome |
| Coverage | Systematic: IEG rates all ICRs; independent rating data available for the full portfolio | Validation sample: IDEV validates a sample of PCRs annually; not all completed projects receive IDEV scrutiny |
| Divergence data | Publicly tracked: IEG vs ICR divergence is published and analysed — the gap is a core accountability metric | Not published as a systematic metric; noted qualitatively in synthesis reports but not tracked as headline data |
| Accountability mechanism | Board can compare management self-ratings against IEG independent ratings project by project | Board sees IDEV synthesis conclusions but not a systematic project-level comparison |
1.2 The Ratings Data, 2015–2024
| Year | PCRs Reviewed | % Satisfactory+ (IDEV) | Key Finding |
|---|---|---|---|
| 2015 | ~70 (est.) | 77% | Baseline figure cited in 2018 synthesis; performance ‘satisfactory’ |
| 2016 | 49 | 77.5% | 38/49 satisfactory+; management PCRs rated ‘significantly higher’ than IDEV PCRENs |
| 2017 | 88 | ~72% (est.) | PCRENs found development effectiveness ‘less than satisfactory on average’; Bank performance ‘systematically rated satisfactory even when project had major implementation issues’ |
| 2018 | 65 | 78% | IDEV 2018 synthesis: 78% (51/65 PCRs) satisfactory; PCR quality average 2.8/4; slight decline from 77.5% in 2016 |
| 2019 | 65 of 89 | 77% | 98% rated relevant — relevance inflation already visible |
| 2020–2021 | 126 | 94% | Unexplained 17-point jump during COVID; 118/126 PCRs positive; no methodology change disclosed |
| 2022–2023 | 130 | 89.2% | 116 of 130 PCRs achieved positive rating. Overall score dropped from 3.15 (2020–2021) to 2.94 — partial correction after the 94% peak. Still well above 2016–2019 average of ~77%. |
1.3 Three Structural Reasons the Numbers Cannot Be Trusted
Validation is not independent rating. The 2017 synthesis makes this explicit: IDEV found that “the Bank’s performance was systematically rated satisfactory or above in the PCRs, even when the project had major implementation issues.” When IDEV subsequently ‘validates’ a PCR that already carries this upward bias, and finds it broadly plausible, it is not correcting the bias — it is certifying it.
The relevance inflation problem. In 2019, 98% of projects were found relevant. A project that set weak objectives, achieved them partially, and completed two years late can score highly on relevance — pulling the composite performance score toward satisfactory even when effectiveness and efficiency are poor. This has been noted repeatedly. It has not been corrected.
Sample selection and the missing failures. IDEV validates a sample of PCRs — not all of them. The selection methodology is not disclosed. If difficult or politically sensitive projects are systematically less likely to be included, the validated satisfactory rate will overstate portfolio performance. The World Bank’s IEG covers all ICRs.
1.4 Adjusted Plausible Performance Range
| Adjustment | Rationale | Implied Rate Adjustment |
|---|---|---|
| Implementation extensions >1 year | 89% of 2018–19 projects extended avg. 2.2 years; proxy for implementation failure | −8 to −12 pp |
| Relevance inflation discount | 98% relevance implausible as a real-world finding | −3 to −5 pp |
| Management self-rating bias correction | 2016 synthesis: management PCRs rated ‘significantly higher’ than IDEV PCRENs | −5 to −10 pp |
| Fragile state underperformance | AfDB portfolio FCS share ~30–35%; no FCS-specific discount in aggregate IDEV ratings | −3 to −6 pp |
| Adjusted Plausible Range | Conservative cumulative adjustment: −19 to −33 points applied to the 94% headline | 61–75% |
Part II — Five Named Projects, Five Satisfactory Ratings
The following cases are drawn directly from AfDB Project Completion Reports — management’s own self-assessments. In each case the PCR text records specific failures. In each case the final rating is Satisfactory. These are not cases IDEV downgraded. These are cases where management wrote the evidence of failure and then awarded itself a passing grade anyway.
Planned: 36 months. Actual: 124 months — over 10 years. The PCR’s own timeliness ratio of 0.29 falls below AfDB’s 0.5 Satisfactory threshold. Rated Highly Unsatisfactory.
Lead contractor Corsan-Corviam was declared bankrupt in 2018, mid-construction. Full restructuring of the joint venture required.
OSBPs at both Senegalese borders were planned core components. Neither was constructed. Senegal’s funding had already closed.
15km of feeder roads planned. Only 3km completed. The remaining 12km simply not delivered.
Employment data for women ‘was not systematically recorded.’ HIV/AIDS, GBV, road safety sensitisation data: ‘not captured — outcome verification impossible.’ The PCR’s own words.
Traffic volumes exceeded the 2014 appraisal target. The OSBPs and feeder roads are treated as scope reductions rather than failures. The timeliness Highly Unsatisfactory sub-rating is arithmetically averaged against a Highly Satisfactory resource use efficiency score. A project that ran 10 years instead of 3, saw its contractor go bankrupt, failed to build two key planned facilities — rated Satisfactory.
The PCR’s exact language: “Project execution and financing is considered very unsatisfactory based on the fact the project took almost twice as long than originally planned to implement.”
The PCR states: “The analysis of cost benefit shows that the Medupi project as installed will not show a positive return and will deliver internal rates of return below the weighted average cost of capital for Eskom.” The Bank financed a project that, by its own model, will not earn back its cost of capital.
Two compliance actions still pending at project closure: Risk to Public Health due to Emissions, and Compliance with Integrated Water Resources Management Policy.
FGD was a planned emission-control component. At PCR it ‘is now planned to be completed by 2027.’ The plant closed without its primary pollution-control system operational.
Resource use efficiency is rated Highly Satisfactory based on 100% physical completion. The PCR then adds: “The quantitative measure used to derive the rating of 4 above is considered to possibly be misrepresentative… qualitatively it is considered that the resource use efficiency was less satisfactory.” The PCR explicitly says its own rating is wrong. It keeps the rating. IDEV validated the PCR.
Loan signed 1 March 2019. First disbursement: 3 December 2019.
Value of working capital provided to women borrowers fell 26% short of target.
Agricultural weather stations were a planned component. Not procured by completion.
“The lack of data on the impact of the agricultural insurance and agricultural banking components on production made it impossible to reconstruct such impact.” Two of three main project components have no measurable economic outcome data.
Development objective rated Satisfactory on the grounds that credit disbursement volumes were achieved. A conditional future tense deployed: “is likely to achieve its development objective provided that BADF improves the rate of credit granted to women and the weather stations are procured.” A conditional future used to justify a present Satisfactory rating on a closed project.
Programme approved for 2020. Not closed until December 2024 — four years after original planned closing.
Disbursement ‘considerably delayed due to failure to meet a disbursement condition relating to the ANRMP’s audit of procurement procedures.’ PARES III funds not disbursed until December 2023 — three years after the programme was originally scheduled to close.
This is a governance reform programme that itself failed a procurement audit condition. The instrument being assessed is governance and accountability reform. The project’s own procurement compliance failure goes unaddressed in the rating rationale.
IDEV found effectiveness satisfactory for outputs ‘but could not provide an assessment on outcomes achievement due to lack of data in the completion report.’
IDEV rated efficiency unsatisfactory due to ‘delays and a relatively low disbursement ratio.’
IDEV ‘did not assess the sustainability criterion, which was not covered in the completion report.’ Not covered. In a completion report. For a six-year regional strategy.
IDEV: “In general, the validation found insufficient evidence to support the completion report’s conclusions that implementation of the RISP has been successful.” IDEV explicitly rejected management’s conclusion. In practice, the management self-rating remains the primary record.
Part III — Seven Techniques for Rating Failure as Satisfactory
Across these five cases and the full IDEV synthesis series (2016–2021), seven techniques consistently convert documented failure into passing grades.
1. Output substitution for outcome. The project counts what was built, lent, or trained — not whether those outputs changed anything. Medupi: 100% physical completion rated despite negative financial returns. PACBA: credit volumes achieved despite no quantifiable economic impact.
2. Scope reduction without penalty. When components cannot be delivered, they are removed from the results framework at mid-term review. The reduced scope becomes the new 100% baseline. Trans-Gambia: 15km of feeder roads planned, 3km built; rated on the 3km.
3. Future-tense satisfaction. The PCR rates a project Satisfactory on the grounds that it ‘is likely to achieve’ its development objective if post-closure conditions are met. A conditional future used to justify a present Satisfactory rating on a closed project.
4. COVID as universal absolution. COVID is cited to explain delays, missed missions, and implementation gaps across the 2020–2021 cohort — the same cohort that achieved a 94% satisfactory rate. If COVID caused the failures, the rating should reflect them.
5. Metric selection to reach the threshold. The composite rating allows high-scoring metrics to offset low-scoring ones. Medupi: resource use efficiency (Highly Satisfactory on fund utilisation) offsets Very Unsatisfactory project execution.
6. Absent data treated as neutral. When outcome data does not exist, PCRs carry the output-level rating forward as a proxy. No data = no failure.
7. Bank performance decoupled from project failure. In all five cases, the Bank’s performance as task manager and supervisor is rated Satisfactory or above — regardless of timeliness ratios, contractor bankruptcy, missed targets, and open environmental violations.
Part IV — What Shareholders Should Ask
- What is the IDEV-rated satisfactory rate broken down by: (a) sector, (b) fragile state vs. non-fragile, (c) instrument type (investment, budget support, TA)?
- What is the divergence between management PCR ratings and IDEV PCREN ratings, project by project, for the 2020–2021 cohort?
- What share of projects in the 2020–2021 cohort received implementation extensions of more than one year? What is their IDEV performance rating distribution?
- On what methodological basis did the satisfactory rate increase from approximately 77% (2016–2019 average) to 94% (2020–2021) during a period of documented implementation stress? Was any methodology change approved by CODE?
- Why does the IDEV synthesis series not publish FCS-disaggregated performance data, given AfDB’s stated commitment to fragile state operations?
- When will IDEV produce a portfolio-wide independent outcome rating series — not PCR validations — comparable to IEG’s PPAR programme?
Annex — Full References
Primary Sources: IDEV Synthesis Reports on PCR Validation
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2016 Project Completion Reports (PCRs). IDEV, 2017. 49 PCRs reviewed; 38 (77.5%) satisfactory or highly satisfactory. Published at idev.afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2017 Project Completion Reports (PCRs). IDEV, 2018. 88 PCRs reviewed; development effectiveness ‘less than satisfactory on average’; Bank performance ‘systematically rated satisfactory or above…even when the project had major implementation issues.’ Published at idev.afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2018 Project Completion Reports (PCRs). IDEV, 2019. 65 PCRs reviewed; 78% (51/65) satisfactory; PCR quality average 2.8/4. Published at idev.afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2019 Project Completion Reports (PCRs). IDEV, 2020. 65 of 89 PCRs reviewed; 77% satisfactory; average score improved to 3.04/4; 98% of projects rated relevant. Published at idev.afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2020–2021 Project Completion Reports and Expanded Supervision Reports. IDEV, 2022. 126 PCRs reviewed; 94% (118/126) achieved positive rating; 16 XSRs reviewed at 93.8% positive. The primary source for the 94% figure. Published at idev.afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). Synthesis Report on the Validation of 2022–2023 Project Completion Reports and Expanded Supervision Reports — Executive Summary. IDEV, 2024. 130 PCRs reviewed; 116 (89.2%) achieved positive rating; overall score dropped from 3.15 (2020–21 cohort) to 2.94. Published at idev.afdb.org.
Primary Sources: Named Project Completion Reports
- African Development Bank. South Africa — Medupi Power Project — Project Completion Report. AfDB Operations Department, 2022. Loan approved November 2009; 4,764 MW coal-fired plant, Lephalale, Limpopo Province. Contains the explicit statement that the Highly Satisfactory resource use efficiency sub-rating is ‘possibly misrepresentative’ and the finding of negative financial return. Published at afdb.org.
- African Development Bank. Trans-Gambia/Senegal — Construction of the Trans-Gambia Bridge and Cross-Border Improvement Project — Project Completion Report. AfDB Operations Department, 2025. Records timeliness ratio of 0.29 (Highly Unsatisfactory); contractor insolvency; OSBPs not constructed; 12km of 15km feeder roads not delivered. Published at afdb.org.
- African Development Bank. Burkina Faso — Support Project for Establishing an Agribusiness Bank (PACBA) — Project Completion Report. AfDB Operations Department, 2022. Records nine-month delay to first disbursement; women’s lending target missed by 26%; weather stations not procured; economic impact of two of three components not quantifiable. Published at afdb.org.
- African Development Bank. Côte d’Ivoire — Economic and Social Reform Support Programme Phase III (PARES III) — Project Completion Report. AfDB Operations Department, 2025. Records four-year delay; disbursement blocked by procurement audit compliance failure; logical framework acknowledged as unrealistic. Published at afdb.org.
- African Development Bank — Independent Development Evaluation (IDEV). West Africa Regional Integration Strategy Paper 2011–2015/17 Completion Report — IDEV Validation Note. IDEV, 2019. IDEV explicitly states it ‘found insufficient evidence to support the completion report’s conclusions that implementation of the RISP has been successful.’ Efficiency rated unsatisfactory; sustainability not assessed. Published at idev.afdb.org.
AfDB Institutional and Results Framework Sources
- African Development Bank. Results Measurement Framework 2016–2025. AfDB, 2016 (updated annually). Contains acknowledgment of ‘questionably high values (90%)’ for the sustainability indicator and commitment to review methodology — the Bank’s own implicit acknowledgment of grade inflation in the sustainability dimension.
- African Development Bank. Annual Development Effectiveness Review 2022. AfDB, 2022. Lending approvals and portfolio composition data; basis for the $8.2bn approvals figure for 2022 and 55% climate finance classification for 2023.
- African Development Bank. Annual Development Effectiveness Review 2023. AfDB, 2023. Records $10.77bn total approvals (UA 8.03bn); 55% classified as climate finance; sovereign operations up 43% year-on-year.
- African Development Bank. ADF-16 Replenishment Report. AfDB, 2022. USD 8.9 billion agreed; largest concessional replenishment in AfDB history.
- African Development Bank. Project Portal Data — South Africa Medupi Power Project (P-ZA-FAA-001). AfDB Projects Portal. Construction commenced May 2008; first unit originally planned for commissioning February 2012. Available at projectsportal.afdb.org.
Independent Institutional Assessments
- Multilateral Organisation Performance Assessment Network (MOPAN). Assessment Report: African Development Bank. MOPAN, 2023. Notes negative trend in NSO evaluation criteria 2018–2020; corroborates concerns about evaluation quality raised in IDEV synthesis reports.
- Universalia Management Group. Quality Assessment of IDEV Evaluation Products. Commissioned by IDEV, 2018. Documents rating scale inconsistencies and quality concerns in IDEV evaluation methodology.
- International Fund for Agricultural Development (IFAD). Annual Report on Results and Impact (ARRI) 2019. IFAD, 2019. Comparative MDB project performance data; provides external benchmark for AfDB figures.
Comparator MDB Evaluation Architecture Sources
- World Bank — Independent Evaluation Group (IEG). Results and Performance of the World Bank Group (RAP) 2023. IEG, 2023. IDA comparator data; basis for the 55–70% Moderately Satisfactory or above figure for comparable IDA operations in Sub-Saharan Africa.
- World Bank — Independent Evaluation Group (IEG). Results and Performance of the World Bank Group (RAP) 2022. IEG, 2022. Confirms IEG independence architecture: IEG rates all ICRs independently; divergence between ICR self-rating and IEG rating published as a systematic metric.
- African Development Bank. Revised Guidelines on Project Completion Report (PCR) Evaluation Note and Project Performance Evaluation Report (PPER). AfDB, 2019. Describes the formal evaluation architecture and the role of IDEV in the PCR validation cycle.
Notes on Sourcing and Methodology
All direct quotations in this note are taken verbatim from publicly available AfDB and IDEV documents as cited above. The adjusted plausible performance range (61–75%) is a structured analytical reframing based on IDEV’s own qualitative findings and is not presented as a statistical estimate. The IDEV-reported satisfactory rate figures for each year have been cross-checked against the published IDEV synthesis reports. The two corrections applied during final verification: (a) the 2018 rate is 78%, not 74% — confirmed from the IDEV 2018 synthesis (51 of 65 PCRs, 78.5%); (b) the 2022–2023 synthesis has been published at 89.2% (116/130 PCRs), not ‘not yet published’ as stated in an earlier draft.
This note has not been commissioned by or submitted to the African Development Bank, IDEV, or any of their member institutions. It reflects the independent analysis of the author.