Horizontal axis is government education spending as a share of GDP. Choose what to plot against it. If money bought results, these would climb left to right.
Ireland, 2.9%
The lowest spender here, and among the highest performers: 41.9% with a degree and PISA reading of 518.
Iceland, 7.3%
The highest spender, at two and a half times Ireland’s share, and below average on both measures.
r = +0.19
Correlation between spending share and reading. For degrees it is +0.15. Both are close to nothing.
99%+
Adult literacy in every country here. The measure stopped discriminating decades ago, which is itself the point.
Share of adults who completed each stage. Primary is finished everywhere. The spread opens at secondary and widens again at higher education, and that is where national differences actually live.
What the chart does not show. There is no meaningful relationship between how much of its economy a rich country devotes to education and what it gets back. Ireland spends 2.9 percent of GDP, the least in this group, and lands near the top on both degrees and reading. Iceland and Sweden both spend 7.3 percent and finish either side of average. Japan spends 3.3 percent and reads at 504. The correlation across all 32 countries is +0.19 for reading and +0.15 for higher education, which is to say the line is nearly flat and the scatter is nearly a cloud.
Two honest caveats. A share of GDP is not an amount of money: Luxembourg’s 3.7 percent and Greece’s 3.4 percent buy very different things per student, because the economies underneath them differ by a factor of several. And spending mix matters enormously, since a country that funds universities publicly reports a higher share than one that pushes the same cost onto households through tuition and debt. The United States is the clearest case: 5.4 percent of GDP in public spending understates total national education expenditure, because a large slice of American higher education is financed privately and therefore never appears in this figure.
The scarcinality reading. Primary attainment is the interesting column precisely because it is boring. Every country here sits between 92 and 100 percent, and adult literacy is above 99 percent across the board. That was not true a century ago. Universal basic education is one of the few genuine scarcities that industrial societies actually abolished, and having abolished it, they stopped being able to distinguish themselves by it. The measure went flat because the problem was solved.
Which is why the money question is the wrong one at this end of the development curve. In a country without schools, spending is the binding constraint and more of it buys enormous gains. In a country that already has universal primary and near-universal secondary, spending is no longer what binds, and the framework’s standing claim applies: when a constraint has been relieved, pouring more of the relieving resource at it produces very little, because the shortage has moved somewhere else. The variation between these countries lives in teacher selection and status, tracking and vocational design, curriculum, and how early the sorting happens, none of which is a budget line. Money is necessary and it is nowhere near sufficient, and the flatness of this chart is what that looks like when you plot it.
Companion to
The Seed Corn Problem, which asks the same question of research funding and finds the composition, not the total, is what moved. Sources: World Bank, government expenditure on education as a share of GDP, and educational attainment for population 25 and over, most recent year available per country, generally 2021 to 2024. OECD PISA reading, 2018 round, the most recent with complete coverage for this group. Attainment measures are the share of adults who have completed each level, a stock measure, rather than annual completion rates, which are unstable and exceed 100 percent for several countries in this group.