French students in fourth grade rank last in the European Union in the 2023 TIMSS evaluations, with 484 points in mathematics. Not second-to-last: last. In a country that invented universal preschool in the nineteenth century and still devotes 5.5% of its GDP to education, this result is not inevitable. It is a consequence.

France ranks significantly below the World Bank’s 2025 Human Capital Index compared to Estonia (0.81) and Singapore (0.90). The gap is not cosmetic: it measures the probability that a child born today will reach their full productive potential in adulthood. Each point lost on this index represents millions of children whose trajectories are set on a downward slope from the start — before first grade, before preschool, before even school policies have had time to act. The problem begins very early. And it is precisely there that public attention turns away.

The Essentials

  • France ranks significantly below Estonia (0.81) and Singapore (0.90) on the World Bank’s 2025 Human Capital Index
  • French fourth-grade students score 484 points on TIMSS 2023, last place in the EU in mathematics
  • Preschool enrollment has been declining for several years, a sign of structural disinvestment according to INSEE and Akademos data
  • Estonia, starting from zero after 1991, built a high-performing education system in thirty years by making early childhood its priority
  • The economic return on early investment is documented in James Heckman’s work: each euro invested before age 5 returns between seven and thirteen euros in cumulative social gains

The Founding Paradox: Invent and Abandon

France created the first asylums for young children in 1826, transformed into preschools under Jules Ferry starting in 1881. For more than a century, universal preschool was one of the strongest arguments in favor of the republican model: admission of children from age 2 or 3, free of charge, social mixing, trained personnel. Countries like Finland, South Korea, and Estonia observed this model and made it the foundation of their educational reforms in the 1990s and 2000s.

Today, these countries are pulling ahead again. Estonia trains its early childhood educators according to curricula revised in 2017, aligned with the latest data from cognitive science research. It enrolls more than 94% of its children aged 3 to 6 in appropriately-sized structures, with adult-to-child ratios among the best in Europe. France, meanwhile, lowered the mandatory school starting age to 3 in 2019, but without increasing the corresponding resources. The legislative gesture was not followed by pedagogical investment.

Preschool enrollments are declining. Positions for ATSEM — the assistants whose role is central to the quality of care for very young children — are under pressure in many municipalities. Initial teacher training for elementary school teachers devotes few specific hours to early childhood pedagogy. The result appears ten years later in TIMSS rankings. The correlation between preschool investment and school results is well established, and the more recent rigorous causal studies point in the same direction. It is simply ignored in budget construction.

What TIMSS 2023 Really Reveals

484 points. This figure deserves to be understood in context. The TIMSS scale in mathematics for fourth grade was calibrated with a central point set at 500 in 1995, and the average of EU and OECD countries in 2023 stands around 524-525 points. Singapore reaches 615 points. The Netherlands, Poland, Hungary, and the Czech Republic all exceed 520. France finds itself behind countries whose per capita income is significantly lower than its own.

This is not a problem of money in the raw sense. France spends more per student than Estonia, measured in purchasing power parity. The question is one of allocation: where and when the money is spent. Research in education economics, notably the work of James Heckman, Nobel Prize laureate in economics, has documented with precision what practitioners observe intuitively: the marginal return of educational investment is maximized between 0 and 5 years, and declines thereafter. Each euro invested in this window generates between seven and thirteen euros in cumulative social gains — health, employment, crime prevented, cohesion. After primary school, this ratio collapses.

France has built a system that does the opposite. It invests massively in high school and higher education — preparatory classes, elite schools, support programs for the baccalaureate — and treats preschool as an adjustment sector. When budgets tighten, it is the primary level where cuts are politically easiest, because parents of preschoolers vote less collectively than associations of high school students and unions in higher education.

The result of this implicit choice, never formulated as such, can be read in social mobility data: France is one of the OECD countries where the correlation between the father’s education level and the son’s is strongest. Not because republican meritocracy is a lie — it produces real effects — but because the dice are loaded before first grade.

What Estonia Did Differently

In 1991, Estonia emerges from Soviet occupation with a centralized, rigid education system ill-suited to a market economy. In thirty years, it builds one of Europe’s most high-performing systems. This is not a miracle: it is the result of a series of deliberate, documented, and replicable choices.

First choice: preschool teacher training is treated as a national priority, with academic requirements comparable to those of primary school. In France, daycare educators and childcare assistants are trained over shorter periods and at levels far inferior to their Nordic or Baltic counterparts. The implicit idea that caring for very young children requires fewer skills than teaching older ones is contradicted by all available cognitive data: it is exactly the opposite.

Second choice: the Estonian preschool curriculum is structured around language development, logical thinking, and emotional regulation — the three most powerful predictors of later school success according to available meta-analyses. This is not about schooling children too early, making them sit at desks at age 3. It is about structuring play, language interactions, exposure to simple reasoning, in environments designed for this purpose.

Third choice: Estonia maintained low adult-to-child ratios despite budget constraints in the 2000s. It considered the quality of individual interaction between the trained adult and child to be non-negotiable. It is this ratio — not buildings or equipment — that determines the essential pedagogical effect in comparative studies.

The comparison also applies to the Netherlands, Poland since 2010, and more recently Portugal, which launched a comprehensive plan to overhaul its preschool system in 2018 and is climbing TIMSS rankings faster than expected. They share in common having treated early childhood education as a measurable and priority investment for the future, not as a third-tier social expense.

The Levers Exist, There Is No Shortage of Examples

The good news — and it is real — is that the action levers are identified, costs are estimated, and timelines for impact are known. This is not about reforming the entire French education system at once: effects would be diluted and resistance would be maximal. It is about acting on three specific points, two of which do not even require immediate spending increases.

The first lever is ongoing training for elementary school teachers assigned to preschool. There exists today in France a corpus of scientific knowledge translated into pedagogical practices — Mireille Brigaudiot’s PARLER Project, Dominique Alamichel’s work on language, guides published by the National Scientific Council for Education on reading acquisition — that remains little disseminated in initial training and nearly absent from ongoing training. Scaling this up does not cost billions: it costs training hours, trained district coordinators, and clear political priority.

The second lever is the recruitment and training of ATSEM. These assistants are present in preschool classrooms and play a daily interaction role with children that education science researchers describe in qualitative data as decisive, particularly for children from linguistically understimulating families. Yet their training is short, their status precarious, and their numbers stagnate while the number of children they supervise increases.

The third lever is welcoming 2-year-old children. France had developed a practice of admitting children under 3 to preschool that constituted a significant innovation and real equal opportunity for children from disadvantaged backgrounds. This rate, which reached 35% in the early 2000s, has fallen below 10%. Available research — particularly evaluations of the Sure Start program in the United Kingdom and follow-up studies of French cohorts from INSEE — confirms that entry at age 2 into a quality structure produces measurable effects on language acquisition at age 5, and these effects are most marked for children whose family environment is least stimulating.

These three levers do not require creating a new institution. They require taking seriously an existing institution that has served as a model for the entire world.

The Missing Political Horizon

What is blocking is not technical. France has the research, the data, comparable examples, and structurally, the resources. What is missing is a political horizon capable of valuing results that are measured ten years after the decision.

The logic of the electoral cycle is known and documented — as are the institutional biases that discourage long-return investments. In education, the effect is particularly severe: a minister who rebuilds preschool in 2025 will never see the results in TIMSS rankings during their term. The minister who made the effort will not reap the political benefit. This is a political externality, in the economic sense of the term: the cost is private, the benefit is collective and deferred.

A few countries have circumvented this problem by institutionalizing educational policy continuity through independent agencies, public five-year objective contracts, and standardized evaluations that make decline visible and political. Finland has its National Agency for Education, independent of ministerial cycles. Estonia has its digital skills objectives set for ten years. These structures do not eliminate political play, but they create a form of accountability that transcends mandates.

In France, the National Scientific Council for Education, created in 2017, plays a role in advice and dissemination of evidence-based data. Its scope and resources remain limited. The TIMSS 2023 results were published, commented on, then absorbed by current events. They did not produce visible budget reorientation.

Investment in early childhood poses the same question as that explored in the debate on vocational training or school guidance policies: do we choose educational sectors based on their demonstrated social return, or based on the political visibility of their beneficiaries? Switzerland chose the first. France still hesitates.

What Can Still Change

France is not condemned to its 484 points. Estonia has progressed notably through successive TIMSS cycles: the progression is slow, it requires ten to fifteen years of consistency, but it is linear once the foundations are laid. Portugal has substantially improved its scores between 2015 and 2023 after a targeted preschool reform. Poland has progressed comparably over the same period, reaching 546 points in 2023.

The World Bank’s Human Capital Index provides, for the first time systematically, a comparative measure that decision-makers can use to arbitrate their priorities. It shows that the gap between France and its most high-performing neighbors is not a definitive chasm: it is a lag constructed over twenty years of relative disinvestment in the early years of life, and it can be closed by applying policies whose effects are documented.

The question is not whether France can do better. It proved this for a century, by inventing the model that the entire world is borrowing from it today. The question is whether the political class is capable of putting back at the top of its agenda a priority whose results will only be visible after the next election, and the one after that.

Sources

  1. Akademos / World Bank HCI 2025: https://www.akademos.app/blog/effondrement-niveau-scolaire-france
  2. TIMSS 2023 — International Association for the Evaluation of Educational Achievement (IEA), results published in December 2023
  3. James Heckman, “Skill Formation and the Economics of Investing in Disadvantaged Children”, Science, vol. 312, 2006
  4. World Bank, Human Capital Index 2025 — https://www.worldbank.org/en/publication/human-capital
  5. National Scientific Council for Education, guides and synthesis notes — https://www.csn.education.gouv.fr
  6. OECD, Education at a Glance 2024, data on per-student spending and early schooling rates
  7. TIMSS 2023 - France Grade 4 Mathematics Score (484 pts)
  8. TIMSS 2023 - Singapore Grade 4 Score (615 pts)
  9. TIMSS 2023 - Poland Grade 4 Score (546 pts)
  10. Law 2019-791 - Mandatory Education at Age 3
  11. OECD EAG 2023 - France 5.5% of GDP (2020)
  12. World Bank Human Capital Index - Singapore 0.90
  13. Heckman - Return on Educational Investment 0-5 Years
  14. Enrollment Rate of 2-Year-Olds - Decline Below 10%
  15. Invention of French Preschool Nineteenth Century
  16. Estonia - Kindergarten Participation Rate (>94%)