FCV Strategy Series · Paper 3 of 6 · MDB Reform Platform
Four Root Causes
Why the World Bank Keeps Underperforming — and the Accountability Architecture That Sustains It
The Multi-Criteria Picture
IEG rates each project across five quality criteria. Across 1,403 evaluated IDA projects with full data, none of the five criteria is met by a majority of projects in either IDA or FCS contexts. Quality at Entry is the weakest — met in fewer than 29 percent of IDA projects — confirming that most delivery failures originate upstream at design, not downstream during implementation.
| Quality Criterion | IDA S+% | FCS S+% | IDA+FCS S+% | Implication |
|---|---|---|---|---|
| Quality at Entry (QaE) | ~42% | 33.3% | 31.2% | Projects designed to be approved, not to succeed |
| Bank Performance (overall) | 39.9% | 30.8% | 29.0% | Under one-third meet overall Bank performance bar |
| Quality of Supervision | 57.0% | 46.2% | 44.5% | Better than QaE but still minority; structural underfunding of field supervision |
| M&E Quality | 41.0% | 41.0% | 41.0% | 80 years of M&E requirements; still below 50%; incentive problem, not technical |
| Development Outcome | 30.6% | 25.8% | 25.8% | One in four IDA+FCS projects achieves a satisfactory development outcome |
The 80/20 Finding
Denizer, Kaufmann and Kraay (2013) — a World Bank Policy Research Working Paper using over 6,000 projects evaluated between 1983 and 2011 — established a finding that has not altered institutional practice: country-level factors explain only 20 percent of the variation in project outcomes. The remaining 80 percent is attributable to Bank-level factors, including the quality of the Task Team Leader. TTL fixed effects are “of comparable importance to country fixed effects” in accounting for outcome variation. The individual TTL’s track record across prior projects is a significant predictor of performance on any given operation.
Four Institutional Pathologies
The IEG evaluation data is consistent with four institutional pathologies identified repeatedly across the RAP series, portfolio assessments, and independent evaluations. These findings recur across a decade of reviews. They are known, well-documented, and the institution has not resolved them.
The Wappenhans Report (1992) identified an ‘approval culture’ — preoccupation with new lending at the expense of implementation quality, driven by incentives that reward portfolio size over performance. The RAP 2024 documents the same pattern thirty-two years later. Quality at Entry — ~42% S+ in IDA — is the weakest of the five criteria precisely because the institutional incentive at the design stage is to produce a project that will be approved, not one that will work. Raballand, Roundell and Mallberg (2015) documented that there is no direct correlation between a poorly performing project and task leader career trajectory. Eighty percent of TTLs report that lending pressure crowds out learning.
Implementation Completion Reports are authored by the project team — typically the TTL — whose career benefits from producing a Moderately Satisfactory rating. The IEG ICRR provides an independent check, but its findings sit in a separate report. For portfolio databases, country strategy papers, Board briefings, and future project appraisals, operations are recorded at the ICR rating, not the ICRR rating. The ICR-ICRR divergence across the full portfolio is substantial: the managed rating propagates through the institutional system; the honest rating does not. The ADB requires recusal from ICRs by evaluators who worked on the project. The World Bank has no equivalent policy.
Quality of Supervision achieved 43.8% S+ in IDA — better than QaE but still a minority. In FCS contexts where field presence is costliest and most critical, supervision budgets are structurally insufficient. The Bank’s Facetime Index — introduced in IDA21 — is a recognition of the problem. It is not yet a solution. Supervision investment per dollar committed in FCS is lower than in non-FCS, despite the operating environment being more complex, projects being more fragile, and the consequences of inadequate oversight being more severe. The incentive structure has not changed: disbursement is rewarded; supervision is a cost.
The World Bank’s resident Board approves approximately 300 operations per year. The 2009 Zedillo Commission named the problem: the Board attempts an impossible trinity of political representation, technical banking supervision, and management oversight. The co-optation consequence is categorical: the Board approved all operations. It held management formally accountable for the development results of none of them. Having approved a project, the Board cannot scrutinise its failure without implicating its own prior endorsement. The institutional response to documented failure is therefore predictable: note the restructuring paper, accept the rating upgrade, continue. The Board is not a passive victim of this dynamic. It approved the accountability architecture that guarantees its own distance from ground-level outcomes.
The Foundation of Impunity: Sovereign Immunity and Zero Consequence
Behind all four root causes lies a structural condition that makes them rational rather than aberrant: the World Bank operates under a framework of sovereign immunity that insulates it — and its staff — from every financial and legal consequence of institutional failure.
Nigeria will repay $387.6 million over 38 years for the Saving One Million Lives programme that IEG rated Moderately Unsatisfactory with efficiency Negligible. The Bank earns its spread on every dollar repaid. A poorly designed $500 million operation and a well-designed $500 million operation produce an identical financial return to the institution. The sovereign guarantee that every IDA borrower provides means that no matter how badly the Bank designs, supervises, or evaluates a programme, the money comes back. This is not incidental to the accountability failure. It is its foundation.
In a commercial lending relationship, the lender bears financial consequences when its own institutional failures contribute to poor outcomes. Due diligence failures, design defects, and supervision failures expose the lender to liability. The World Bank bears none of these consequences. Its Articles of Agreement grant it immunity from legal process in member country courts. Its staff are protected by functional immunity for acts performed in their official capacity. No TTL has faced personal financial consequence for a project rated Unsatisfactory. No Practice Manager has faced institutional consequence for approving a Quality at Entry that IEG later documents as inadequate. No Country Director has had compensation reduced because the portfolio they managed produced 20 percent Satisfactory outcomes.
The consequences of this structure fall entirely on the borrowing country. Nigeria repays $387.6 million regardless of outcome. The IDA borrower in South Sudan, DRC, or Haiti repays the credit over 38 years regardless of whether a single intended beneficiary was served. The Bank, meanwhile, records the credit on its balance sheet as a performing asset, earns its net income, and uses that income to fund the next lending cycle. The asymmetry is total: the institution that designs the operation bears no downside; the country that requested the operation bears all of it.
This structural immunity propagates through the institution at every level. Staff know — not through cynicism but through experience — that the career consequence of raising concerns that impede approval is negative, and that the career consequence of approving a programme that subsequently fails is zero. The institutional term for a specialist who raises concerns that block a project is “roadblock.” There is no institutional term for a TTL whose project is rated Unsatisfactory five years after approval, because there is no institutional consequence to name.
Reforming this requires changes at two levels. At the institutional level: outcome-linked partial debt relief when IEG documents that Bank institutional failures materially contributed to poor outcomes — funded from Bank net income, not from IDA resources — so that the institution bears a defined share of the financial cost of its own design failures. At the staff level: IEG assessments documenting specific institutional failures must be formally recorded in performance files, and operations with such assessments must be ineligible as positive portfolio evidence in promotion decisions for the relevant TTL, Practice Manager, and Country Director for a defined period. Paper 6 sets out both recommendations in specific, actionable terms. Neither requires amending the Articles of Agreement. Both can be implemented through Board resolution.
Paper 2: $70 Billion Below Standard — The IDA Delivery Record 2015–2025
Paper 3: Four Root Causes — Why the Bank Keeps Underperforming · Current
Paper 4: IFC in Fragile States — 11 Percent Satisfactory
Paper 5: IDA21 — What It Gets Right and What It Doesn’t Fix
Paper 6: Eight Recommendations in Priority Order