Learning Outcomes
Universal schooling has nearly arrived; universal learning has not. PISA 2025, published in September 2026, recorded the lowest OECD averages yet in mathematics and reading (the chart below shows the 2022 cycle). The World Bank estimates 70% of 10-year-olds in low- and middle-income countries cannot read and understand a simple paragraph — a measure of 'learning poverty'. Across PISA-participating economies, the variation in 15-year-old reading and maths skills is wide and only loosely correlated with spending.
Key insights
Schooling is necessary but not sufficient for learning
Two children in formal schooling for the same number of years can learn vastly different amounts depending on quality. Vietnamese 15-year-olds have scored close to the OECD average in PISA despite far lower per-pupil spending and a shorter history of universal schooling. Several oil-rich Gulf countries spend more per pupil than the OECD average but score well below. Inputs (spending, teachers, hours) explain less variation than the quality of pedagogy, accountability and parental support.
The pandemic widened the learning gap
School closures during 2020–22 affected most of the world's children. World Bank estimates: average learning loss of 0.5–1.0 years globally; greater for disadvantaged children. Recovery is uneven — high-resource countries have largely closed the gap; many low-income countries have not. The cohort that lost 18–24 months of effective schooling has measurable consequences for lifetime earnings and skill formation.
East Asia leads, consistently
Singapore, China (B-S-J-Z province samples), Hong Kong, Macao, Taiwan, South Korea and Japan top PISA reading, maths and science rankings. The pattern is robust over decades. Common features: strong teacher selectivity, intensive after-school tutoring culture (the 'shadow education' system), cultural emphasis on academic achievement, and high accountability frameworks. Replicating outcomes outside East Asia has proven hard.
PISA 2022 average score — selected economies
Mean of mathematics, reading, science (OECD avg = 478)
Show data table
| Average score | |
|---|---|
| Singapore | 560 |
| Macao | 535 |
| Taiwan | 533 |
| Japan | 533 |
| South Korea | 523 |
| Estonia | 516 |
| Switzerland | 498 |
| Canada | 506 |
| Finland | 495 |
| Germany | 482 |
| USA | 489 |
| OECD avg | 478 |
| Spain | 477 |
| Italy | 477 |
| Indonesia | 374 |
| Brazil | 397 |
| Philippines | 353 |
Key Finding: East Asia leads; OECD average sits near 478; many emerging markets sit well below 400.
Learning poverty by region
% of 10-year-olds unable to read simple text
Show data table
| % LP | |
|---|---|
| High income | 8 |
| Europe/Central Asia | 13 |
| East Asia/Pacific | 21 |
| Latin America | 52 |
| Middle East/N.Africa | 63 |
| South Asia | 60 |
| Sub-Saharan Africa | 86 |
| Low income (all) | 89 |
Key Finding: Pre-pandemic learning poverty was over 80% in Sub-Saharan Africa and under 10% in advanced economies.
Methodology & caveats
PISA, TIMSS, PIRLS
PISA tests 15-year-olds across OECD and partner economies every 3 years (81 countries and economies in 2022); covers reading, maths, science. TIMSS tests 4th and 8th graders in maths and science (~60 countries). PIRLS tests 4th grade reading. PISA samples by age; TIMSS by grade. PISA emphasises real-world application; TIMSS emphasises curriculum content. The three measure different things and don't always rank countries the same.
Learning poverty measure
World Bank Learning Poverty = % of 10-year-olds unable to read and understand a simple text. It combines two components: % out of school + % in school but below minimum proficiency. The threshold uses harmonised scores from regional assessments (LLECE in Latin America, PASEC and SACMEQ in Africa, etc.). Measure introduced in 2019; the launch estimate was 53%, the 2019 update 57%, and the simulated post-pandemic estimate is 70% in low- and middle-income countries.
Why direct comparisons are hard
Cross-country test comparisons require: harmonised test instruments, equivalent translations, representative samples, and statistical equating across years and languages. PISA invests heavily in this; smaller regional assessments often don't. Caveats are particularly important for trends (years of changing test items) and for emerging markets with weak sampling frames.
Cite this page
GlobeChart. “Learning Outcomes.” Last updated 2026-10-31. https://globechart.com/education/learning-outcomes/
Please also credit the original publisher named under each chart: OECD PISA 2022; World Bank Learning Poverty Brief 2022. Spotted an error? Tell us.