A Finn, an Icelander and a Shanghainese Walk into a PISA Test

In the early 2000s, I often travelled to Espoo in Finland on work. I was impressed by the country: its efficiency, its transition from a minerals and forest economy to a knowledge economy, and the competence of the people I encountered. Finland was also the darling of comparative education.

Its students performed exceptionally well in PISA, the OECD's assessment of 15-year-olds across countries and education systems. Unlike a conventional examination, PISA is designed to test how well students can apply knowledge and skills to unfamiliar problems. It is a particularly interesting measure of what an education system might be achieving.

For a time, Finland was the country every educationist wanted to visit. Its schools seemed to offer an appealing alternative to the more high-pressure Asian systems. Finnish teachers became objects of international fascination.

And, inevitably, explanations followed. Finland succeeded because of teacher autonomy. Because there was less testing. Because children played more. Because the system was equitable. Because Finnish society valued education. Some or all of these may have mattered. But once we have a successful outcome, it is easy to assemble an explanation around it.

The recently published PISA 2025 results complicate that story. Finland is still above the OECD average, but its performance has fallen substantially. Between 2015 and 2025, its scores declined across science, mathematics and reading.

The point is not that Finland has suddenly become a bad education system. It hasn't. The point is that the same system that once appeared to contain some of the secrets of educational success now looks less exceptional.

What happened? Did the features we once identified as the sources of Finland's success disappear? Did something else change? Was their relationship with educational performance dependent on circumstances that have themselves changed? Or did we simply attribute too much explanatory power to factors that happened to coexist with a period of exceptional performance?

Then there is Iceland.

Iceland is a different kind of problem. It is not a poor country that has poor PISA results. By most broader measures of human development, it is an extraordinarily successful society: wealthy, highly educated, socially developed and possessing a very high standard of living.

Yet between 2015 and 2025, Iceland's PISA scores fell by 37 points in science, 46 points in mathematics and an extraordinary 64 points in reading.

So what happened? Again, some of the obvious explanations don't fit. Icelandic students report above-OECD levels of perseverance, curiosity, goal setting, help seeking, cognitive adaptability and growth mindset. Family support actually increased between 2022 and 2025, when it declined in almost every OECD country. Iceland also has relatively strong digital preparedness, including substantial exposure to learning how to evaluate AI-generated information.

There is an especially intriguing finding in the report. Iceland performs well in computational problem solving once its science performance is taken into account. Its difficulties in the traditional PISA domains therefore do not translate neatly into a general story about an inability to reason.

So we have a country whose teenagers report positive attitudes towards learning, whose schools have strong digital capabilities, whose family support has recently improved, and whose students display strength in computational problem solving. Yet its performance in the core PISA domains has fallen dramatically.

Which brings us to China.

The Chinese jurisdictions of Beijing, Shanghai, Jiangsu and Zhejiang, grouped together by PISA as B-S-J-Z, are extraordinary performers, topping the rankings with scores of 597 in science and 612 in mathematics. In reading they are second only to Singapore.

This is exactly the kind of result that invites easy explanations. Curriculum. Teacher quality. Parental expectations. Academic culture. Discipline. Selection. Time spent studying. Social expectations. Take almost any feature associated with the Chinese education system and it can be incorporated into a plausible account of its performance. But this is where we should pause.

Imagine trying to understand why a particular restaurant is among the best in the world. We could study the menu, examine the ingredients, interview the chef and identify the techniques used in the kitchen. We might then produce a list of the characteristics of excellent restaurants and send aspiring restaurateurs off to copy them.

But a restaurant is not just its ingredients or techniques. The chef matters, but so do the suppliers, the kitchen brigade, the local food culture, the history of the cuisine, the economics of the restaurant, the setting and the expectations of its diners. Some things can be copied. Others cannot. And even when two restaurants use similar ingredients and techniques, they may produce very different meals.

Education systems are similar. We identify a successful system and start hunting for its recipe. China gives us an extraordinary outcome and a multitude of plausible explanations. Finland gives us a successful outcome that has worsened. Iceland gives us a large decline that is difficult to explain. In each case, we can construct a story. But is the story the actual explanation?

The outcomes of education emerge over many years from enormous numbers of interacting variables. We cannot take China's education system, transplant it into another country and expect it to work. But this is not an argument against comparing education systems. Quite the opposite. Looking carefully at places where something remarkable is happening is one of the great strengths of comparative education.

The problem begins when looking becomes searching, searching becomes explanation, and explanation becomes prescription. Finding excellence is easier than explaining excellence. And explaining excellence is easier than knowing how to reproduce it.

PISA 2025 is actually a very useful report. It is much richer than the league tables that dominate the headlines. The report examines not only proficiency but engagement, belonging, teacher and family support, school climate, socioeconomic inequality, digital distraction, AI literacy and other dimensions of students' lives and learning.

Yet the public conversation continues to compress all of this into a handful of numbers. China is high. Singapore is high. Finland has fallen. Iceland has fallen further. The ranking is useful only to tell us where something interesting is happening.

But instead of using PISA as a search for the world’s smartest kids or a hunt for the education system that has solved schooling, it might be more insightful to use it as a vast source of questions about how learning works. The real prize in comparative education may not be finding the world's best education system but developing hypotheses that can be tested across time and across cultures.

And India? We had a look once. In 2009, Tamil Nadu and Himachal Pradesh participated in PISA. We didn't like the result, so we left the classroom. India has not participated since.