Are We Asking the Right Questions About Education?

KEY TAKEAWAYS

• In France, as in other OECD countries, student performance has seen a steep decline in mathematics and reading.

• The latest Programme for International Student Assessment (PISA) shows that this decline affects disadvantaged and advantaged students alike.

• International comparisons are more useful for questioning education systems’ capacity to reform than for identifying a model to copy.

The publication of the latest results from the Programme for International Student Assessment (PISA 2025) on September 8 prompted the usual flurry of commentary, comparisons and rankings. France has slipped, some countries have improved, while others have proved more resilient. As always, attention quickly turns to where each country stands.

These rankings are useful, but they risk obscuring a more important question: what do we actually make of the years we spend in school?

For much of the 20th century, measuring educational progress seemed relatively straightforward. We counted students, years of schooling and qualifications. There were good reasons for doing so: the expansion and democratisation of education were among the major social transformations of the past century.

But now that the vast majority of children spend many years in school, the question has shifted. It is no longer enough to know how long they remain in education. We need to understand what we make of those years, and what those years enable us to become. This is precisely the purpose of the work we have undertaken with our colleague Nadir Altinok in Edumetrics: Measuring Human Capital for the 21st century. By reconstructing comparable measures of the quality and distribution of learning over the long run, we seek to shift the analysis of human capital away from years of schooling towards the skills actually acquired.

Rankings aside, how far have we come?

The PISA 2025 results bring this question into sharper focus. Across OECD countries, average performance is now at its lowest level since PISA began in all three major domains assessed. Between 2015 and 2025, scores fell by 22 points in mathematics and 28 points in reading.

Twenty per cent of 15-year-olds are now low performers in all three main domains, compared with 16% in 2022.

France scores 483 points in science, almost exactly the OECD average of 482. It performs slightly below the OECD average in reading (456 compared with 461) and mathematics (458 compared with 463). But perhaps what matters more than France’s position is where it’s heading: since 2022, it has lost a further 18 points in reading and 16 points in mathematics.

Whether a country ranks 25th, 27th or 30th ultimately tells us less than knowing where it has come from, which direction it is moving in, and which students are progressing or falling behind.

This is also suggested by a recent report from France’s High Commission for Strategy and Planning on educational attainment. It adds an important qualification to the usual diagnosis. France suffers from substantial educational inequalities, but its difficulties do not concern disadvantaged students alone. They also affect advantaged students and, in some areas, even the highest performers.

There is therefore both a problem of inequality and, at least up to lower secondary school, a problem of relatively low overall attainment, particularly in mathematics and science. Looking only at averages is not enough; neither is looking only at inequalities.

Expanding access does not yet mean democratising learning

France has dramatically expanded access to education. This is a historic achievement that should not be underestimated. But when access becomes almost universal, democratisation must also become qualitative. The question is no longer simply: who gets into school? It becomes: what do students learn there, and who learns what?

Our long-run series shows that, after a period of international convergence in educational performance until around 1990, national trajectories began to diverge again. Since 2000, relatively few countries have experienced a sustained improvement in the quality of learning. PISA 2025 does not, of course, bring this history to an end. It simply adds another observation to it.

This perspective also allows us to look differently at educational inequalities. When potential skills are not developed, it is not only individual trajectories that are affected: part of a society’s collective human capital is unable to ever fully develop.

Our simulations for France suggest that economic growth could have been around 0.5% higher if an effective policy to reduce educational inequalities had been implemented since the late 1970s.

Spending more is not enough

Resources obviously matter. The High Commission highlights several weaknesses in France: relatively modest expenditure on primary education, a comparatively high number of students per teacher by European standards, and growing difficulties in recruiting teachers.

Resources matter, but they do not explain everything. PISA 2025 suggests that the relevance of policies and the way resources are allocated appear to be more closely associated with performance than spending per student alone. This is also one of the lessons of Edumetrics: spending more is not enough; we need to invest better.

Yet between one euro spent on education and learning actually acquired lies an entire world: teachers, students, families, curricula, teaching practices, schools and institutions, and time. Resources do not mechanically produce outcomes. They have to be transformed into capabilities.

This leads to a question we ask less often: students learn, but do education systems themselves know how to learn?

A reform produces intended effects and others that are less predictable. However, too many reforms can defeat the purpose of reform itself. A practice may work in one context and fail in another. Policies need to be adjusted, developed further and sometimes abandoned. Education policy is, therefore, also about a collective ability to observe, experiment, measure, correct and learn.

From this perspective, international comparisons are valuable, not in so much for telling us what to copy but in allowing us to ask better questions.

And then came artificial intelligence

This question takes on a new dimension with artificial intelligence (AI). In a matter of seconds, a student can now obtain a translation, an explanation, a summary, the solution to an exercise or an apparently convincing piece of writing.

One might conclude that some forms of foundational knowledge are becoming less important. Yet PISA highlights that some of the reading skills now in decline are precisely those we need in an information-saturated environment: evaluating information, comparing multiple sources, interpreting evidence and exercising critical judgement.

Human capital in the 21st century can probably no longer be conceived as a stock of knowledge acquired once and for all. It must also encompass a more fundamental capability: learning how to learn. Initial education remains essential because it provides the foundations that subsequently enable people to revise their knowledge, acquire new skills and adapt throughout their lives.

Measure to understand, not to rank

One essential caveat must nevertheless remain: tests are not education, and PISA is not school. No indicator can capture everything a child learns while growing up. Data can inform our decisions; they cannot decide for us what we wish to pass on or what we collectively wish to become.

This is also why the latest PISA findings should probably not be read as evidence of inevitable decline. Some learning journeys are improving. The United Kingdom and Türkiye, for example, have made progress in science.

There is, of course, no single model that can simply be replicated. History never provides recipes; sometimes it offers possibilities. PISA therefore invites us, perhaps, to ask our questions about education in a slightly different order.

We ask how much we spend, how many students succeed, which countries are moving forward and which are falling behind. All these questions are legitimate. But perhaps they come after another one: what does an education system actually manage to transform?

Time into learning, knowledge into skills, skills into the ability to adapt and, perhaps, these individual capabilities into collective possibilities. That is what we should seek to measure instead of reducing education to a handful of numbers, in order to distinguish the illusion of progress from the reality of progress as far as possible.