Public-school share alone does not explain secondary completion or gender gaps

Countries with more public-school enrollment do not consistently post higher completion or smaller gender gaps. When comparable official evidence is placed side by side, funding arrangements, education pathways and family background repeatedly separate outcomes more clearly than provider share alone. [26] [8] [45] [46] [47]

Key findings

63%, 88% — Public-school share does not map cleanly to completion. Chile has a private secondary share of roughly 63%, yet 88% of upper-secondary entrants complete within two years of the theoretical graduation date. [32] [45]

8%, +11 pp — A large public-school share does not guarantee a small gender gap. Italy’s private secondary share is about 8%, yet women’s completion exceeds men’s by 11 percentage points in the OECD cohort measure. [32] [47]

17%, 83%, ~0 pp — Finland has a private secondary share of roughly 17%, an 83% completion rate within two years, and almost no gender difference, showing that public-heavy systems can still produce very different gender patterns. [32] [46]

92%, >80% — About 92% of Dutch upper-secondary students attend privately managed institutions, but more than 80% of those institutions are government-dependent. Private management is not the same as private financing. [8] [48]

78%, 72% — Across participating OECD systems, on-time upper-secondary completion in 2024 averaged 78% for women and 72% for men. Analyses that assume girls are always the disadvantaged group miss much of the observed pattern. [26]

96.1%, 1% — Korea combines 96.1% public funding across primary to non-tertiary secondary education with only 1% of young adults below upper-secondary attainment, but attainment and cohort completion are different outcomes and should not be treated as the same metric. [2] [3]

First separate public management from public funding

The first mistake in a cross-country comparison is to collapse every system into only two boxes: public and private. In UIS and OECD classifications, a private institution is defined by who controls and manages it. A privately managed school becomes government-dependent when at least half of its core funding comes from government or its teachers are paid by government. A high private-management share therefore does not automatically imply high household costs or weak public responsibility. [18] [15] [41] [42] [8]

That distinction changes the research question. Asking whether public-school enrollment predicts completion is not the same as asking whether public funding is adequate, whether private schools are regulated, or how much households pay. The Netherlands makes the difference unusually clear, and it shows why two countries with similar private shares can generate very different outcomes when funding and regulation differ. [18] [15] [41] [42] [8]

Ignoring this distinction automatically labels systems with high private management as marketized. A safer dataset keeps management, financing, fees, admissions and accountability in separate columns so government-dependent and independent private institutions are not treated as equivalent. [18] [15] [41] [42] [8]

Private secondary enrollment rose, but the rise itself does not explain outcomes

The GEM Report estimates that the global share of secondary students in private institutions rose from about 19% in 2004 to 26% in 2014 and then broadly stabilized. Across OECD upper-secondary systems, the private-institution share also moved from roughly 18% in 2013 to 23% in 2022. These are different universes and should not be spliced into one series, but both show that non-state management has become more prominent over time. [10] [13] [11] [40]

The important point is that this does not necessarily mean public financing retreated. Government-subsidized religious, nonprofit, foundation and vocational institutions can expand private management while public funding remains high. A time-series analysis therefore needs two separate variables: who runs the institution and who pays for it. [10] [13] [11] [40]

The long-run trend documents a larger role for non-state providers, but the reasons differ by country. Historical religious-school settlements, vocational provision, urban growth, shortages of public places and school-choice reforms can all increase the private share, so the trend itself should not be read as the cause of completion outcomes. [10] [13] [11] [40]

Public-school share and completion do not move in one direction

Even a small set of comparable cases fails to produce a simple slope. Chile has a private secondary share of roughly 63%, yet 88% of upper-secondary entrants complete within two years of the theoretical graduation date. Finland combines a private share of about 17% with 83% completion, while Italy combines about 8% private enrollment with 84% completion. Completion does not mechanically rise as the private share falls. [32] [45] [46] [47] [29]

This is not a three-country regression and should not be presented as one. Its value is different: a strong claim that “more public provision means better completion” does not survive basic counterexamples. Student composition, vocational pathways, tracking, repetition rules, family income, teacher allocation and financing all operate alongside provider type. [32] [45] [46] [47] [29]

The selected-case scatter is not a causal estimate, but it is useful for rejecting a monotonic story. A larger panel should match reference years and then test whether any slope survives controls for income, urbanization, inequality and resources per student. [32] [45] [46] [47] [29]

Gender gaps do not always disadvantage girls

Across participating OECD systems, on-time upper-secondary completion in 2024 averaged 78% for women and 72% for men. Among 25-34 year-olds, the share below upper-secondary attainment was 14% for men and 11% for women. In many higher-income systems, male under-completion is now at least as important as female disadvantage. [26] [1] [9] [30]

Extending that pattern to the whole world would be another mistake. Nepal’s estimated ultimate upper-secondary completion is 44% for women and 53% for men, the opposite direction, while lower-secondary completion in 2024 was 89% for both sexes. The direction of the gap changes by country and level, so an equity analysis needs to ask not only how large the gap is but who is disadvantaged. [26] [1] [9] [30]

A gender-gap measure should retain its sign rather than only its absolute size. Using women minus men by education level distinguishes male under-completion in many high-income systems from female disadvantage in some lower-income contexts. [26] [1] [9] [30]

Chile: high private enrollment can coexist with high completion

Chile disrupts a simple public-versus-private story. UIS-linked data put the private secondary share at roughly 63% in 2024, while the OECD reports that 88% of upper-secondary entrants complete within two years of the theoretical graduation date. Women’s completion probability is six percentage points higher than men’s. [32] [45] [6]

Those numbers do not prove that private schools caused the result. Chile combines school choice with subsidies and regulation, and its system also faces socioeconomic segregation. The case therefore shows the weakness of provider share as a stand-alone predictor, not the superiority of one provider type. [32] [45] [6]

Chile is informative only if the institutional design behind the private share is kept visible. Subsidies and regulation matter, and raw achievement differences between schools mix school effects with student selection. That selection problem also matters when drawing conclusions from national aggregates. [32] [45] [6]

Finland and Italy: public-heavy systems can have different gender outcomes

Finland and Italy are both relatively public-heavy in secondary education, yet their gender patterns differ. Finland has a private share of roughly 17%, an 83% completion rate within two years and almost no gender gap. Italy has about 8% private enrollment and 84% completion, but women lead men by 11 percentage points. [32] [46] [47] [26]

The shared characteristic of a large public sector cannot explain the different gender outcomes. Vocational versus general pathways, flexibility between tracks, regional inequality, repetition, dropout and family background can produce different results inside systems that look similar on provider share alone. [32] [46] [47] [26]

The Finland-Italy contrast suggests that who manages the school may be less important for gender equity than where boys and girls leave, repeat or delay along the education pathway. Track mobility, vocational design and student support can be more direct policy levers than provider share. [32] [46] [47] [26]

The Netherlands: 92% private management does not mean weak public finance

The Netherlands shows how misleading the word “private” can be. About 92% of upper-secondary students attend privately managed institutions, yet more than 80% of those institutions are government-dependent. Across primary to non-tertiary secondary education, 88.2% of funding is public, and private schools have long had a right to public financing on terms comparable with public schools. [8] [48] [4] [11]

Calling the Dutch system simply “market-oriented” because of its private-management share misses the institutional core. Legal management is non-state, while financing and rules remain strongly public. Cross-country work therefore needs private-management share and public-funding share side by side. [8] [48] [4] [11]

In the Netherlands, private management is compatible with a strongly public financing model. The case argues for measuring publicness through universal access, funding responsibility, fee rules and accountability rather than legal ownership alone. [8] [48] [4] [11]

Nepal: sector composition and completion gaps can point in different directions

Nepal shows why sector composition cannot be treated as an outcome. In upper-secondary enrollment for 2024/25, girls represent 52% of students in public community schools but 42% in private institutional schools. Yet estimated ultimate upper-secondary completion is 44% for women and 53% for men, a nine-point female disadvantage. [9] [6] [29]

At lower secondary, completion in 2024 is 89% for both sexes. The gender pattern changes sharply by education level within the same country. Sector composition, transitions between levels, dropout and household costs need to be separated before asking whether public schools are more equitable. [9] [6] [29]

Nepal also highlights timing. Students enroll, transition and complete over several years, so a sector composition measured in one year should not be paired mechanically with completion in the same year as if it were an immediate cause-and-effect relationship. [9] [6] [29]

Korea: high public funding and high attainment are informative but not the same metric

Korea is a case where strong public financing and high long-run attainment coexist. Government provides 96.1% of funding across primary to non-tertiary secondary education, government spending per student is USD 21,476 in PPP terms, and only 1% of 25-34 year-olds are below upper-secondary attainment. [2] [3] [14]

Those numbers should not be ranked directly against the cohort-completion measures for Chile, Finland and Italy. Adult attainment is a cumulative outcome across cohorts and uses a different denominator from whether current entrants graduate on schedule. Korea shows that high public financing can coexist with high attainment, but it does not prove causality on its own. [2] [3] [14]

Korea’s combination of high public funding and high attainment is important, but household education spending and an exam-intensive progression system also shape outcomes. Public funding is therefore a meaningful explanatory variable without being a complete explanation on its own. [2] [3] [14]

Education spending must be measured separately from provider share

Average government spending per upper-secondary student across the OECD is USD 13,174 in PPP terms. General programmes average USD 12,421 and vocational programmes USD 13,901, illustrating how programme structure changes costs. Resources per student are a separate axis from public-school share, and similar provider mixes can support very different staffing, facilities, vocational equipment and special-needs provision. [27] [39] [31] [34]

Government expenditure as a share of GDP is not enough either. Countries with different income levels and student populations can devote the same GDP share but deliver very different resources per learner. High per-student spending can also reflect higher wages and prices. A useful model therefore needs PPP spending per student, expenditure relative to GDP, and the split between public and household financing. [27] [39] [31] [34]

The quality of finance depends on where money goes, not only where it comes from. Spending on teachers, counselors, vocational equipment, transport, grants and digital infrastructure can have very different effects on persistence and student experience even at the same per-student level. [27] [39] [31] [34]

Ignoring income, urbanization and inequality exaggerates provider effects

Richer countries can invest more in teachers, transport, digital infrastructure and student support. Urbanization can shorten access distances and simplify staffing, while also creating concentrated poverty and residential segregation. Inequality measures such as the Gini index capture how family background can vary even when national average income looks similar. [36] [37] [38] [29] [31]

A simple correlation between provider share and completion can easily misread these background conditions as a school-sector effect. At minimum, PPP GDP per capita, urban population share, inequality and spending per student should be pre-specified controls, with sensitivity checks that separate OECD/high-income systems from lower- and middle-income systems. [36] [37] [38] [29] [31]

More controls are not always better. When the number of complete-country observations falls, an overloaded model becomes unstable. Pre-specifying a small set of theoretically necessary controls and moving other variables to sensitivity models is more transparent. [36] [37] [38] [29] [31]

Teacher allocation and class structure can turn similar spending into different outcomes

Across the OECD, student-teacher ratios are around 13 to one in lower and upper secondary education, and average lower-secondary class size is 23. National averages, however, conceal regional and subject differences. Rural versus urban schools, general versus vocational programmes and special-needs provision can convert similar spending per student into very different levels of actual support. [28] [27] [44]

That is why the relationship between education spending and completion is not mechanically positive. Extra resources may reduce teacher shortages in one country but mostly absorb higher wages or maintenance costs in another. Allocation across students and programmes can be a more direct mechanism than whether the provider is legally public or private. [28] [27] [44]

Student-teacher ratios are not a perfect measure of instructional quality either. Teacher experience, subject specialization, vacancies and actual teaching hours differ. Even so, staffing is closer to the resources students experience than a legal public/private label. [28] [27] [44]

Matching the same year sharply reduces the usable sample

A practical weakness of international education data is that the latest reference year differs by indicator. Private secondary enrollment is available for 2024 in some countries but only 2018-2023 in others, while government education expenditure can refer to anything from 2022 to 2025. Completion estimates also vary in the amount and quality of underlying observation. [32] [33] [34] [35]

If each country’s “latest” value is pasted into one row, systems observed in different policy and economic conditions are treated as if they belonged to the same year. The primary analysis should prioritize same-country, same-year complete cases and explicitly accept the smaller sample. Mixed-latest-year comparisons are safer as sensitivity checks. [32] [33] [34] [35]

Missingness is unlikely to be random. Countries with weaker statistical capacity, conflict or fragility may be more likely to lack recent completion and finance data, so complete-case samples can tilt toward more stable systems. The composition of the analytical sample should be reported as a result in its own right. [32] [33] [34] [35]

Since 2015, international goals have focused on completion, equity and quality rather than provider share

SDG 4.1 calls for all girls and boys to complete free, equitable and quality primary and secondary education, while SDG 4.5 targets gender and other disparities. The Incheon Education 2030 framework treats education as a right and public good while emphasizing inclusion, equity, gender equality, adequate public finance and accountability. [23] [24] [25] [29]

Provider share is not the target variable in that international framework. Private institutions can exist inside a system designed to meet public goals through funding, regulation and accountability. Conversely, a predominantly public system can still miss equity and completion goals if household costs, regional gaps, teacher shortages or dropout remain high. [23] [24] [25] [29]

For policy, the core test is not who owns the school but whether all students can complete without prohibitive cost, whether disadvantaged groups receive additional support, and whether public finance is tied to accountability. Those principles can apply to both public and private providers. [23] [24] [25] [29]

Institutional combinations are more informative than a public-versus-private effect

Putting the cases together reveals at least four institutional combinations: public management with public finance, private management with public finance, private management with mixed finance, and more independent private provision with larger household burdens. The Netherlands is the clearest example of the second type; Chile is closer to a subsidy-and-choice mixed model; Korea shows that mixed management can coexist with very high public funding. [8] [3] [40] [41] [44]

The classification matters because two systems with the same “50% private” share can expose students to very different costs and rules. Admissions, fees, teacher employment, subsidies, accountability and vocational pathways need to be coded alongside provider share if the analysis is meant to explain real institutional mechanisms. [8] [3] [40] [41] [44]

The classification is not designed to rank countries. Placing systems on management and funding axes makes it easier to ask why countries with similar private shares produce different outcomes and clarifies what a later regression or case study needs to control. [8] [3] [40] [41] [44]

Counterevidence and alternative interpretations

First, a high private share does not necessarily imply weak equity. The Netherlands combines dominant private management with public financing and comparable fee rules, while Chile combines a high private share with relatively high cohort completion. These cases weaken interpretations that assign outcomes to private management itself. [8] [48] [45]

Second, a large public share does not guarantee a small gender gap. Italy’s low private share and +11-point female advantage, together with Nepal’s gender pattern changing across education levels, show that other mechanisms intervene between provider structure and equity. [47] [9]

Third, higher average performance in private schools can reflect household income, parental education and residential selection. In lower-income settings, fees and private education costs can also increase saving and borrowing burdens, so socioeconomic selection must be controlled before provider effects are interpreted. [6] [29]

Implications for policy and analysis

Policy dashboards should track upper-secondary completion, dropout, the direction of gender gaps, public resources per student and household costs rather than treating the public-school share as a target in itself. Provider type is better used as one institutional explanatory variable among several. [18] [27] [28] [36] [37] [38] [41]

Countries with large private sectors should separate government-dependent from independent private institutions. Whether publicly financed private schools follow comparable rules on access, fees, admissions and accountability determines what the private share actually means. [18] [27] [28] [36] [37] [38] [41]

Cross-country research should not merge every country’s latest observation by default. A same-country, same-year complete-case sample is a stronger primary design even if it is smaller, because it reduces spurious correlations created by asynchronous reference years. [18] [27] [28] [36] [37] [38] [41]

Gender equity metrics should retain both the sign and the education level of the gap. Male under-completion in many high-income systems and female under-completion in some lower-income systems require different policy responses. [18] [27] [28] [36] [37] [38] [41]

When expanding education finance, governments should publish not only expenditure as a share of GDP but also PPP spending per student, staffing, vocational costs, targeted support and household spending so resource allocation can be linked to completion outcomes. [18] [27] [28] [36] [37] [38] [41]

Conclusion: does a larger public-school share improve completion and gender equity

What we know with confidence

What we can say with confidence is that public-school enrollment share alone does not consistently explain upper-secondary completion or gender gaps. Completion and gender differences in Chile, Finland and Italy do not line up with the ranking of their private shares, and the Netherlands combines extremely high private management with predominantly public finance. Cross-country analysis is mis-specified if management and financing are not separated. [8] [45] [46] [47]

What depends on conditions

The extent to which public finance or public management supports completion depends on resources per student, household fees, vocational pathways, teacher allocation, national income, urbanization and inequality. The same funding share can produce different results depending on where the money goes, and the same private share means something different in government-dependent and independent private systems. [27] [28] [36] [37] [38] [41]

What remains unknown

What remains unknown is the independent causal effect of public management after income, urbanization, inequality and resources are controlled in a same-year country panel. Reference years are asynchronous and government-dependent private categories are missing in many countries, so a single effect size would not yet be robust without a sufficiently large complete-case sample and stronger microdata. [32] [33] [34] [14]

Final answer to the research question

The final answer is closer to “no.” A larger public-school share by itself does not reliably predict higher secondary completion or a smaller gender gap. The more consistent explanation separates school management from public finance and then adds resources per student, household costs, income, urbanization, inequality and education pathways. Both public-heavy and private-heavy systems can produce strong or weak outcomes; what matters more is how the system actually allocates funding, access and support. [26] [8] [45] [47] [41]

Sources and method

The analysis prioritizes official or public evidence from UNESCO UIS, OECD, UIS-linked World Bank Data, the GEM Report and the United Nations. Metrics with different definitions were not pooled into one outcome, and only directly interpretable values were used in tables and figures. When case values refer to different years, they are treated as counterexamples and institutional comparisons rather than causal estimates. [18] [14] [15] [16] [17] [33]

• [18] UNESCO Institute for Statistics: UNESCO UIS Data Browser Glossary
• [26] OECD: Education at a Glance 2026 — Upper secondary transitions and outcomes
• [27] OECD: Education at a Glance 2026 — Financing upper secondary education
• [28] OECD: Education at a Glance 2026 — Student-teacher ratios and class sizes
• [29] UNESCO Global Education Monitoring Report: 2026 Global Education Monitoring Report — Access and equity
• [31] UNESCO UIS / GEM Report / World Bank: 2024 Education Finance Watch
• [23] United Nations: Transforming our world: the 2030 Agenda for Sustainable Development
• [24] UNESCO: Education 2030: Incheon Declaration and Framework for Action
• [10] UNESCO Global Education Monitoring Report: 2021/2 GEM Report — Non-state actors in education
• [11] OECD: OECD Review of Digital Education Policy in the Netherlands — Governance
• [13] OECD: Education at a Glance 2024 — Upper secondary system features
• [1] OECD: Education at a Glance 2026 — Educational attainment by gender
• [3] OECD: Education at a Glance 2025: Korea
• [8] OECD: OECD Review of Digital Education Policy in the Netherlands — Infrastructure, funding and procurement
• [9] UNESCO Global Education Monitoring Report: Nepal — 2026 GEM Report country case study
• [14] OECD: Education at a Glance 2026 — Sources, Methodologies and Technical Notes
• [32] World Bank Data / UNESCO UIS: School enrollment, secondary, private (% of total secondary)
• [33] UNESCO / UIS / GEM Report: Completion rate estimates
• [36] World Bank Data: GDP per capita, PPP (current international $)
• [37] World Bank Data / UN Population Division: Urban population (% of total population)
• [38] World Bank Poverty and Inequality Platform: Gini index
• [41] OECD: Education at a Glance 2026 — Key system-level finance definitions
• [45] OECD: Education at a Glance 2026: Chile
• [46] OECD: Education at a Glance 2026: Finland
• [47] OECD: Education at a Glance 2026: Italy
• [48] OECD: OECD Review of Digital Education Policy in the Netherlands