Quality at Entry in IDA · Part 14 of the Zero Club Series · MDB Reform Platform
The Variable IDA Controls — and Does Not: When both design quality and M&E are strong, 73% of IDA projects achieve Satisfactory outcomes. When both are weak — which is the case in 52% of the portfolio — 5.4% do.
The Independent Evaluation Group (IEG) is the World Bank’s independent evaluation arm. It reports directly to the Board of Executive Directors — not to management — and reviews every World Bank project after it closes. For each project, the World Bank’s own team prepares an Implementation Completion and Results Report (ICR), rating its own performance. IEG then produces an independent Implementation Completion and Results Report Review (ICRR): a desk-based critical validation of that self-assessment, drawing on the ICR, the original Project Appraisal Document, the Financing Agreement, and the Country Partnership Framework. Before finalising its review, IEG’s evaluator interviews the last Task Team Leader. A senior panel reviewer then reviews the draft. The ICR team can submit written comments. IEG adjusts its ratings if the new information is material. This is not a cursory check. It is a structured, multi-step independent audit process with a formal challenge mechanism built in.
For Investment Project Financing (IPF) — the standard World Bank investment loan — IEG evaluates Bank Performance along two dimensions: Quality at Entry and Quality of Supervision. Quality at Entry (QAE) is a retrospective rating of how well the Bank designed and appraised the project before it was presented to the Board for approval. According to IEG’s Guidance Manual for Validators (IPF, last revised May 2024), the rating covers ten criteria: (1) strategic relevance — whether the project objectives were aligned with country priorities and the Bank’s own Country Partnership Framework; (2) technical, financial, and economic analysis — whether the design was technically feasible, the economic analysis credible, and the cost estimate realistic; (3) poverty, gender, and social aspects; (4) environmental aspects, including safeguard provisions; (5) fiduciary aspects — financial management and procurement systems; (6) policy and institutional aspects — whether the policy environment and government capacity were correctly assessed and the theory of change was credible; (7) implementation arrangements — whether the project was assigned to an agency capable of delivering it, with realistic coordination across ministries and a functioning Project Management Unit; (8) monitoring and evaluation design — whether measurable outcome indicators, baselines, and targets were in place before disbursement; (9) risk assessment — whether principal risks were identified, quantified where possible, and matched with realistic mitigation plans; and (10) Bank inputs and processes — whether adequate preparation resources were deployed, lessons from prior comparable projects were applied, and the preparation timeline was sufficient to produce a sound design.
For Development Policy Financing (DPF) — the policy-conditionality budget support loan used heavily by the MTI practice — IEG evaluates Bank Performance differently. The DPF Guidance Manual (last revised April 2024) does not assign a standalone “Quality at Entry” rating. Instead, IEG rates Bank Performance through the quality and credibility of the prior actions: whether each condition required of the government before disbursement was substantive, addressed a real reform constraint, and was connected through a credible results chain to the stated objective. This distinction matters for interpreting the data in this paper. The QAE ratings analysed here are drawn primarily from IPF projects. DPF operations receive a Bank Performance rating on different criteria, which partly explains why DPF portfolios can record high Bank Performance ratings even when development outcomes are poor: the evaluation framework assesses whether the government completed the required prior action, not whether the prior action was sufficient to produce the development change the operation claimed.
The critical implication for the QAE analysis: every QAE rating in this paper represents IEG’s independent retrospective judgment on decisions the Bank made before disbursement — decisions over which the Bank had complete control. When IEG rates a project Moderately Unsatisfactory on Quality at Entry, it is documenting that the Bank’s own preparation, appraisal, and design fell below an adequate standard before a single dollar was disbursed. The rating is not a claim about what the country did during implementation. It is a claim about what the Bank chose to design and what the Board chose to approve.
This paper analyses a variable the Zero Club series has not yet examined directly: the quality of IDA’s own project design. Two variables — Quality at Entry in IDA and M&E Quality — are entirely within IDA’s control. When both are strong, 73 percent of IDA projects achieve Satisfactory outcomes. When both are weak — which is the case in 52 percent of the portfolio since FY2000 — 5.4 percent do. IDA controls both. It does not consistently require either to meet a minimum standard before Board approval.
When QAE is Satisfactory, 63 percent of IDA projects achieve Satisfactory outcomes. When QAE is Moderately Unsatisfactory, 1.3 percent do. The gradient persists within every institutional quality band — in the weakest countries and the strongest. Country difficulty is not the explanation. Design quality is.
The most powerful historical finding is not the cross-section. It is the trend. In the 1980s, a poorly-designed project had a 10.7 percent chance of recovery through supervision. In the 2020s: 0.0 percent. The Bank cannot rescue poorly designed projects. The only remedy is to prevent approving them. 735 IDA projects with poor design quality committed $50.6 billion at a 3.8 percent S+ rate. IDA’s own quality system flagged every one before the Board voted.
QAE gradient table (all 6 tiers). Recovery rate collapse by decade. QAE × M&E interaction (the four quadrants). QAE by Global Practice (Transport worst at 12.8%; PforR best at 65.4%). QAE by instrument and decade. Largest projects with poor QAE ($6.5bn, 14 projects, zero Satisfactory). CPIA robustness tests (3 tests, within-band gradients). Speed-quality trade-off. Lesson analysis: what produces good QAE (results-based financing 2.6x, community participation 1.9x). IEG evaluator text analysis (136 projects, 9 failure themes). The approval culture: Wappenhans to the present.
The Central Finding: The QAE Cliff
IEG rates every evaluated IDA project on Quality at Entry using a six-tier scale. Across 2,141 IDA projects in Sub-Saharan Africa since FY1980, the relationship between design quality and outcomes is not gradual. It is a cliff.
Source: IEG ICRR/PPAR database, March 2026. Africa only. FY1980–2026. IDA projects only.
The critical threshold is between Satisfactory (63.0%) and Moderately Satisfactory (17.4%) — a 46 percentage-point drop. Once QAE falls below Satisfactory, the probability of a Satisfactory outcome collapses. At Moderately Unsatisfactory, 1.3 percent: one project in 77. At Highly Unsatisfactory: zero.
The Recovery Collapse: Why Supervision Cannot Compensate
The most powerful finding in the historical data is not the cross-tabulation. It is the trend in the recovery rate — the probability that a poorly-designed project (QAE below Satisfactory) achieves Satisfactory outcomes through good supervision or adaptive management.
| Decade | QAE = Satisfactory → S+ Rate | Poor QAE → S+ Rate (Recovery) | Direction |
|---|---|---|---|
| 1980s | 57.8% | 10.7% | Some recovery possible |
| 1990s | 60.4% | 8.9% | Declining |
| 2000s | 63.6% | 1.6% | Near-zero |
| 2010s | 67.3% | 0.5% | Effectively zero |
| 2020s | 71.0% | 0.0% | Zero — no rescue possible |
The Bank has gotten better at executing well-designed projects: 57.8 percent S+ in the 1980s, rising to 71.0 percent in the 2020s. But the recovery rate from poor design has collapsed from 10.7 percent to zero over the same period. Three structural changes explain it: projects have become more complex and harder to restructure mid-course; the Bank is accelerating preparation timelines from 19 to 12 months while IEG’s RAP 2023 documents that preparation quality has declined from 59 to 54 percent; and project ambition has escalated in precisely the settings where institutional capacity is weakest.
The policy implication follows directly. The Bank cannot rescue poorly designed projects. It can only prevent approving them. The $3.575 billion FY2026 institutional budget, against projected lending of $79 billion — a real decrease of 0.7 percent in administrative resources while lending volume rose 58 percent — is not calibrated for quality. The IDA21 $300 million Grant Facility for Project Preparation funds better preparation. It does not prevent Board approval of projects with a 1.3 percent probability of success.
The Interaction: QAE and M&E Together
Quality at Entry and M&E Quality — the two variables most directly within IDA’s control — interact powerfully. When both are strong, 72.8 percent of IDA projects achieve Satisfactory outcomes. When both are weak — which is the case in 52 percent of IDA’s Africa portfolio since FY2000 — 5.4 percent do.
| QAE & M&E Combination | Projects | S+ Rate | Portfolio Share |
|---|---|---|---|
| S+ QAE + Substantial/High M&E | 316 | 72.8% | 22% |
| S+ QAE + Weak M&E | 78 | 47.4% | 6% |
| Poor QAE + Substantial/High M&E | 263 | 25.9% | 19% |
| Poor QAE + Weak M&E | 740 | 5.4% | 52% |
The diagonal tells the story of Bank capability. Good design compensates partially for weak M&E (47.4%). Strong M&E compensates partially for weak design (25.9%). But when both fail — which is the majority of IDA’s Africa portfolio — the outcome is near-certain failure. $62.8 billion has been committed to the bottom-right quadrant since FY2000. IDA controls both variables. It does not consistently require adequate quality in either before Board approval.
Design Quality Is Not a Proxy for Country Difficulty
The most obvious objection to this paper’s central finding: QAE may simply reflect country quality. Harder countries receive lower QAE ratings and worse outcomes. If so, QAE is capturing country conditions, not Bank design decisions.
Three tests address this directly. The key test: within each CPIA band — holding institutional quality roughly constant — does QAE still predict outcomes? It does, with near-identical gradients in every band.
| CPIA Band | QAE = Satisfactory | QAE = Mod. Unsatisfactory or worse | Gradient |
|---|---|---|---|
| Below 3.0 (weakest) | 65.5% S+ | 0.8% S+ | 64.7pp spread |
| 3.0–3.5 | 61.8% S+ | 1.6% S+ | 60.2pp spread |
| Above 3.5 (strongest) | 71.3% S+ | 1.1% S+ | 70.2pp spread |
In the weakest institutional environments, a well-designed IDA project achieves 65.5 percent S+. A poorly-designed project achieves 0.8 percent. Country difficulty shifts the level slightly — the strongest environments produce marginally better outcomes for well-designed projects — but the 60–70 percentage-point QAE gradient dwarfs the CPIA effect. This is not a country problem. It is a design decision problem.
QAE by Global Practice and Instrument
By Global Practice: Transport worst, SPJ best
| Global Practice | QAE S+ Rate | Outcome S+ Rate | Gap |
|---|---|---|---|
| Transport | 12.8% | 11.6% | 1.2pp |
| MTI | 31.5% | 19.5% | 12.0pp |
| FCI | 29.6% | 24.1% | 5.6pp |
| Health | 25.4% | 23.9% | 1.4pp |
| Education | 33.1% | 27.3% | 5.8pp |
| Agriculture | 24.5% | 28.7% | –4.2pp |
| Social Protection & Jobs | 47.0% | 54.0% | –7.0pp |
Transport has the worst QAE in Africa: only 1 in 8 transport projects meets the design standard. MTI carries the widest QAE-to-outcome gap: 12 percentage points, reflecting that even when DPF projects are adequately designed, the instrument’s structural weaknesses produce an execution gap. Social Protection & Jobs is the only GP where outcomes exceed QAE — its direct transfer model has structural features that compensate for weak entry quality. Finance, Competitiveness and Innovation, the GP that owns the MSME agenda at the centre of the FCV Strategy, records 29.6 percent QAE Satisfactory: seven in ten FCI projects go to the Board below design standard.
By Instrument: PforR forces quality by design
| Instrument | Projects | QAE S+ Rate | Outcome S+ Rate | Why |
|---|---|---|---|---|
| PforR | 26 | 65.4% | 61.5% | Requires Disbursement-Linked Indicators with verified measurement before approval. Quality assurance is structural. |
| DPF | 393 | 33.1% | 22.1% | Prior actions verifiable on paper without functional implementation. Conditionality rewards legal compliance. |
| IPF | 1,224 | 29.1% | 28.5% | No structural quality requirement. QAE depends entirely on preparation incentives. |
PforR achieves 65.4 percent QAE Satisfactory not because PforR teams are more conscientious but because the instrument’s design requires measurable objectives, verified results frameworks, and independent DLI verification before Board approval. The instrument builds quality assurance into its structure. DPF’s 11-point QAE-to-outcome gap confirms that even adequate DPF design does not translate into sustained reform, because conditionality that can be met by letter-verification systematically overstates reform progress.
What IEG’s Own Evaluators Say
IEG evaluators write explanatory text when they rate a project’s QAE below Satisfactory. Across 136 IDA Africa projects rated Moderately Unsatisfactory or below on QAE, their explanations follow a consistent pattern:
| Design Failure Theme | Frequency | What This Means |
|---|---|---|
| Institutional capacity mismatch | 74% | Designed for capacity that does not exist. The most common single failure. The design team knew — or should have known — before Board approval. |
| Inadequate preparation quality | 72% | Insufficient analysis before Board presentation. The preparation timeline or resource allocation was not adequate to produce a sound design. |
| Risk assessment failure | 70% | Risks underestimated or mitigation plans absent. The design acknowledged the environment but did not calibrate ambition to it. |
| M&E / results framework weakness | 56% | Indicators that cannot measure outcomes, or baselines that do not exist. The project cannot demonstrate what it achieves. |
| Excessive design complexity | 56% | Too many components, too many ministries, too many objectives for the operating environment. The single strongest predictor of failure in lesson text analysis. |
| Lessons from prior projects not applied | 53% | IEG documented the same lesson in a prior project. The current design repeated it. Institutional amnesia operating at project level. |
These are not retrospective inferences. They are IEG’s own quality assessors explaining, in real time at evaluation, why the design they reviewed was inadequate. The quality assessment was available within the approval process when each project was presented to the Board.
The Speed-Quality Trade-off
The Bank is actively reducing project preparation time. President Banga announced in October 2024 that preparation had been cut from an average of 19 months to 16, with a target of 12 months. The Bank’s internal OEE Dashboard tracks preparation time as a priority indicator “with the directional arrow pointing toward shorter times.”
IEG’s RAP 2023 reported a decline in the share of investment projects with high preparation work quality from approximately 59 percent to 54 percent over the period in which preparation was accelerated. The Bank is moving faster. The quality is getting worse. The FY2026 institutional budget is $3.575 billion for projected lending of $79 billion — a real decrease of 0.7 percent in administrative budget against a 58 percent increase in lending volume. Fewer resources per dollar committed to ensure adequate design. The recovery rate has fallen to zero. The direction of travel is not ambiguous.
The Counterfactual: What Better Design Would Produce
The reform is straightforward in principle. If no project with below-Satisfactory QAE were approved, IDA’s Africa S+ rate would rise from approximately 30 percent to approximately 63 percent. The Bank would lend approximately 30–40 percent fewer projects — and achieve more. Fewer projects means fewer IDA credits burdening poor countries’ debt portfolios for 38 years. More successful outcomes means more development impact per dollar committed.
The Wappenhans Report proposed this in 1992. No subsequent institutional reform has implemented it. The incentive structure has not changed. The recovery rate has fallen to zero. The case for the reform is stronger than when it was first proposed thirty-four years ago.
The Board: 53 Years of Reports, Zero Corrective Action
Every project in this record was approved by the Board of Executive Directors. The Board owns the outcome risk. IEG has been transmitting independent evaluations of those outcomes to the Board for 53 years — since its predecessor began systematic evaluation in 1973. Across more than half a century, the Board has taken no formal corrective action in response to the persistent below-Satisfactory pattern those evaluations document. The reports are received, noted, and shelved. The poor results keep being delivered.
This makes IEG, in its present institutional position, largely cosmetic. Not because IEG fails its mandate — it produces rigorous independent evaluation — but because its findings reach a Board that co-approved every project in the record and therefore cannot scrutinise that record without implicating its own decisions. Management receives IEG’s findings first and frames the Board’s encounter with them. The evaluation exists; the consequence does not. The situation the Bank is in is, ultimately, the Board’s responsibility.
What We Are Proposing
The reform is not more analysis. The Bank’s own data already identifies the problem with precision. What is required is a change to the institutional machinery that approves projects the data show will fail:
A project that the Bank’s own evaluators would rate below Satisfactory at entry should not proceed to the Board. The standard already exists; it is simply not binding.
Restore the independent pre-Board design review that the Quality Assurance Group provided until it was disbanded in 2014. Nothing replaced it. This is the single highest-leverage institutional change available.
A QAE threshold below which a project cannot be presented to the Board. Quality at entry should function as a hard constraint on approval, not a rating recorded after the fact.
Across every Global Practice, staff are rewarded for moving lending to Board approval, not for whether projects achieve their development objectives. Link career outcomes to IEG outcome ratings so that delivering results, not approving volume, is what advances a career.
Stop approving projects the data show are designed to fail. After 53 years of IEG evaluation and zero corrective action, the Board must convert documented failure into consequences — for management, and for itself.
Series Context
The Zero Club Series (Parts 9–13)
43 countries, 14 Global Practices, 419 projects, $34.5bn, zero Satisfactory. This paper adds the institutional cause: IDA approved these projects with below-Satisfactory design quality. The Zero Club is the outcome; poor QAE is the mechanism.
Sound Strategy, Broken Platform (FCV Assessment)
The companion FCV strategy assessment. The strategy introduces a $300M Grant Facility for preparation. It does not create a mechanism that prevents approval of projects with 1.3% S+ probability. The QAE paper documents what that mechanism must address.
Game Theory: Why the System Does Not Learn
The QAE gradient is the data. The game theory paper explains the mechanism. TTL payoffs reward approvals, not outcomes. Poor QAE at evaluation carries no career consequence. The approval culture is a Nash equilibrium.
The Largest Poor-QAE Projects
14 projects, $6.5bn combined, all with MU or worse QAE at approval. Tanzania health ($1.06bn, MU, MU). Tanzania roads ($791M, HU, HU). DRC MTI ($450M, MU, U). Kenya transport ($300M, U, U). Not one achieved Satisfactory.
| # | Paper | Commitment | S+ Rate | Status |
|---|---|---|---|---|
| 9 | Zero Club — MTI in Africa | $10.4bn | 0% | Published |
| 10 | Zero Club — Health in Africa | $3.0bn | 0% | Published |
| 11 | Zero Club — Transport in Africa | $1.65bn | 0% | Published |
| 12 | Zero Club — Education in Africa | $2.5bn | 0% | Published |
| 13 | Zero Club — Cross-Sector Synthesis | $34.5bn | 0% | Published |
| 14 | Quality at Entry in IDA (this paper) | $175bn | 3.8% (poor-QAE portfolio) | This paper |
QAE gradient (all 6 tiers, 2,141 projects). Recovery collapse by decade. QAE × M&E four quadrants. CPIA robustness (3 tests). QAE by GP and instrument. 14 largest poor-QAE projects ($6.5bn, zero S+). Speed-quality trade-off. Lesson analysis. Evaluator text (136 projects). The counterfactual. The approval culture: Wappenhans to the present.
IDA controls two variables that together predict the majority of project outcomes: Quality at Entry and M&E Quality. When both are strong, three in four IDA projects achieve Satisfactory outcomes. When both are weak — 52 percent of IDA’s Africa portfolio since FY2000 — one in eighteen does. The Bank cannot rescue poorly designed projects; the recovery rate is now zero. It can only prevent approving them.
It does not, because the incentive architecture rewards approval over outcomes across every Global Practice, and because the Board that approves every project and receives every IEG evaluation has, in 53 years, never converted that evidence into consequences. The reform is not more analysis. It is restoring Quality at Entry as a binding standard, rebuilding independent quality assurance, shifting internal incentives from pushing money to delivering results, and holding the Board and management accountable for outcomes. The evidence is in. The choice is institutional.