The Zero Club (Part 12) · Education in Africa · MDB Reform Platform
Education in Africa: 10 Countries, 33 Projects, $2.5 Billion Committed, Zero Satisfactory
The Education Zero Club is not 10 separate country failures. It is one institutional pattern: the Bank’s education portfolio measures inputs and outputs — schools built, enrolment up, teachers trained — not learning. IEG’s own 2024 flagship found that projects with learning indicators receive lower ratings than those without. The Moderately Satisfactory equilibrium absorbs the gap between schooling and learning into “partial achievement” and renders it invisible in corporate reporting. But this paper goes further than diagnosis. Kenya, Tusome, LEAP, GEQIP, and the IEG flagship converge on what works: learning improves when the intervention reaches the classroom — through structured pedagogy, teacher coaching, accountability, and continuous measurement.
The Accountability Gap
The World Bank’s Education Global Practice committed $12.9 billion across 207 IEG-evaluated projects in Sub-Saharan Africa. The MS+ rate — counting Moderately Satisfactory as success — is 70 percent. The S+ rate, the benchmark the Bank applies to IFC and MIGA, is 30 percent. The 41-point gap is among the widest of any sector in Africa, exceeded only by Transport (44.4pp) and MTI (42.7pp).
IEG’s 2024 flagship evaluation, Confronting the Learning Crisis, explains why: “Education sector projects are among the best performing of all World Bank projects, but the achievement of their objectives is typically defined and measured in terms of outputs (e.g., number of teachers trained) and not by learning outcomes.” Projects are rated successful because they measure enrolment and inputs. Learning is not consistently measured.
The Anchor Case: Ethiopia
Ethiopia is not a fragile state. It is one of Africa’s fastest-growing economies, a non-fragile lower-middle-income country with a functioning Ministry of Education, and the Bank’s largest education borrower in Africa.
| P-Code | Project | FY | Rating | Commit |
|---|---|---|---|---|
| P000721 | Education VII | FY1988 | U | $70M |
| P000732 | Education Sector Development | FY1998 | MS | $234M |
| P069083 | Ethiopia Distance Learning LIL | FY2001 | MU | $3M |
| P078692 | Post Secondary Education | FY2005 | MU | $40M |
| P106855 | General Education Quality Improvement (GEQIP 1) | FY2009 | MS | $413M |
| P129828 | General Education Quality Improvement (GEQIP 2) | FY2014 | MS | $530M |
| P163608 | Education Results Based Financing | FY2017 | MS | $30M |
| P174206 | COVID-19 Education Response | FY2021 | MS | $15M |
Eight projects. $1.3 billion. 33 years. Ratings: 5 Moderately Satisfactory, 2 Moderately Unsatisfactory, 1 Unsatisfactory. Never once Satisfactory. The flagship GEQIP programme — $943 million across two phases, the largest education engagement in Africa — rated MS twice.
The Independent Evidence
Four independent sources confirm the Zero Club findings through different methodologies. None draws on the IEG project ratings database used in this paper. All reach the same conclusion.
HCI+ (2026)
“Limited progress in learning across nearly all income categories.” In most low-income countries, learning is the same as or worse than 2010. Four in five African 10-year-olds cannot read a simple story.
IEG Learning Crisis (2024)
“Projects measure outputs, not learning outcomes.” Only 22 of 188 operations with teacher training tracked impact on classroom practice. Operations with learning indicators receive lower ratings.
Rolleston et al. (2025)
“Rising access and falling outcomes.” Little evidence of improvement in average learning outcomes under GEQIP, 2012–2021, despite modest gains in teacher quality indicators.
UNESCO GEM (2026)
“Learning outcome levels in mathematics declined.” Rural students who benefited from GEQIP made greater progress, but aggregate scores fell as access expanded to more disadvantaged children.
What Works: Four Models
The evidence from across the Africa portfolio — and from independent evaluation — now supports a simple synthesis. Four models have been tried. Each teaches a distinct lesson, and together they point to the same conclusion:
| Model | Example | Result |
|---|---|---|
| Access expansion | Ethiopia GEQIP | Access rose sharply; learning mixed to flat. Strong systems necessary but not sufficient. |
| Governance reform | Kenya SWAp (FY2007) | Failed (U, $468M). Sector-wide financing without accountability did not deliver. |
| Structured pedagogy | Kenya Tusome | Strong literacy gains. Changing what teachers did in classrooms produced learning. |
| External accountability | Liberia LEAP | Significant gains (0.18 SD), but cost and access trade-offs limit scalability. |
The pattern across all four models is consistent: inputs and systems alone do not move learning. Learning improves when the intervention reaches the classroom — through structured pedagogy, teacher coaching, accountability for results, and continuous measurement.
The Kenya Evolution
Kenya’s education portfolio — eight IEG-evaluated projects across three decades — traces the shift. Generation 1 (1990s–2000s): inputs and institutions. Universities (U), Early Childhood Development (U). The question was: can institutions function? Generation 2 (mid-2000s): governance and accountability. Free Primary Education ($50M, S) worked with controls; the $468M SWAp (U) failed without them. Generation 3 (2015 onward): learning-focused reform. PRIEDE ($88M, MS), KSEIP ($200M, MS), and the COVID response (S) increasingly focus on structured pedagogy, teacher coaching, learning assessment, and results-based financing. Kenya’s Tusome programme achieved some of the strongest literacy gains in Africa within government systems.
The Liberia LEAP Experiment
Liberia’s LEAP programme contracted eight private operators — including Bridge International Academies, Rising Academies, and BRAC — to manage 93 public primary schools. A randomised controlled trial found learning gains of 0.18 standard deviations. But the trade-offs were severe: Bridge expelled teachers and displaced students; costs reached $640 per student versus the government’s $50; access declined. The IFC had invested $13.5 million in Bridge; its Compliance Advisor Ombudsman found “substantial concerns,” and IFC divested in 2022. LEAP supports the accountability hypothesis. It does not support outsourcing as a scalable solution. Kenya’s Tusome achieved comparable gains within government systems, at government cost, without the trade-offs.
The Five Failure Modes
IEG lesson text across all 207 education projects was manually coded. M&E weakness appears 195 times — the highest frequency of any theme in any sector in the Zero Club series. The five failure modes, documented in the Bank’s own evaluation language:
| Failure Mode | Countries | IEG Language |
|---|---|---|
| Design complexity exceeding capacity | Ethiopia, Kenya, DRC | “Civil works are by themselves insufficient to improve the quality of instruction.” (Ethiopia, FY1988, U) |
| Access-quality trade-off unresolved | Zambia, Ethiopia | “Government campaigns to increase access should be accompanied by a commitment to improve learning outcomes.” (Zambia, FY1999, MU) |
| M&E without learning measurement | Burundi, Chad | “Realistic project ratings might have alerted management earlier.” (Chad, FY2003, U) |
| Teacher training without quality assessment | Ethiopia, Cameroon | Only 22 of 188 operations tracked the impact of teacher training on classroom practice. (IEG, 2024) |
| Institutional absence in fragile states | DRC, CAR, Liberia, Congo-Brazza | “Performance-based financing can be an effective instrument in fragile settings.” But viable produced MS, not S. (DRC, FY2017, MS) |
The Cross-Sector Pattern
DRC now appears in five Zero Club lists simultaneously — Health, MTI, Transport, Energy, and Education. Five sectors, five instruments, billions committed, zero Satisfactory in any. Cameroon appears in both Health and Education — the two human-capital investment sectors. The pattern is not sector-specific. It is institutional.
Where Education Works
The Zero Club argument is strengthened by acknowledging where Education succeeds — and why:
| Country | S+ Rate | Commit | Why It Worked |
|---|---|---|---|
| Rwanda | 75.4% | $245M | Imihigo performance contracts at school level; community accountability for learning |
| Mauritania | 66.2% | $179M | Focused literacy targets with explicit learning outcome indicators |
| Cabo Verde | 63.6% | $28M | Small scale; single implementing ministry; genuine government ownership |
| Sudan | 56.3% | $175M | Emergency operations with bounded objectives and external implementation support |
Education succeeds where learning outcome indicators are explicit rather than proxied by inputs; institutional accountability connects teachers and schools to results; project scale matches implementation capacity; and government owns the learning objective rather than the enrolment target. Rwanda demonstrates that Zero Club status is not permanent. Countries can exit when implementation discipline changes.
The Case Study Series
| # | Case | Commitment | S+ Rate | Status |
|---|---|---|---|---|
| 1 | Nigeria Water | $1.8bn | 0.4% | Published |
| 2 | Angola DPF | $2.2bn | 0% | Published |
| 3 | South Africa Energy (Eskom) | $9.13bn | — | Published |
| 4 | Ghana FCI | ~$500M | 0% | Published |
| 5 | DRC Portfolio | $6.7bn | 6.1% | Published |
| 6 | DRC Inga | $107M+ | — | Published |
| 7 | Somalia | ~$900M | 89% | Published |
| 8 | Rwanda | $4.6bn | 68.5% | Published |
| 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% | This paper |
The Education Zero Club documents the fourth sector of sustained underperformance in the World Bank’s Africa portfolio. Ten countries, 33 projects, $2.5 billion committed, zero percent Satisfactory. The 41-point gap between the Bank’s reported success rate and the honest benchmark is among the widest of any sector in the series.
But this paper goes further than diagnosis. Four independent sources — the HCI+, IEG’s Confronting the Learning Crisis, Rolleston et al., and UNESCO — confirm the findings: the Bank’s education portfolio measures inputs, not learning. And the Kenya evolution, Tusome, and LEAP show the pathway out: structured pedagogy, teacher coaching, accountability, and continuous learning measurement.
The binding constraint may now be located inside classrooms rather than inside education ministries. The Bank can get African children into school. Whether they are learning inside it is the question the next decade of lending must answer.