What Makes a Good School System? Comparing 39 Countries

Published 3 October 2026. Updated 6 October 2026 to add related reading links.

Two things made me start looking at school systems.

The first was a conversation with a Danish father living in England. His children were in state schools and wanted to move back to Denmark. His description of their experience was not particularly encouraging, and the obvious alternative in London, private education, was extraordinarily expensive.

A few days later, French students started blocking their schools. The movement had begun at a lycée in Créteil and spread first around Paris and then across France. The students complained about missing teachers and staff, overcrowded classes, deteriorating buildings, heavy timetables and the pressure surrounding university admissions. By the end of September, hundreds of schools had been affected.

Neither of these things told me whether England or France actually had a bad school system, and anecdotes rarely tell us very much on their own. But together they made me curious about a fairly simple question: has anyone actually figured out how to build a good school system?

It turns out to be a much larger question than it sounds. We normally compare countries through PISA and other international tests, and the discussion tends to revolve around who is best at mathematics, who has moved up, why Finland has fallen and why the East Asian countries do so well. These are reasonable questions, but a school system does much more than produce test scores, and it is built for the society around it. The Danish school system reflects Danish ideas about equality, authority, childhood and the role of the state, and the Japanese and French systems reflect their own cultures in the same way.

In that sense, schools are also rather good portraits of the societies that built them. They show what a country expects from children, teachers and parents, how much inequality it accepts, how it thinks authority should work and how much of childhood it believes should be organised around preparation for adult life.

This makes the idea of ranking school systems somewhat questionable. There may not be one ideal system that can be copied from one country to another, because a system has to work inside a particular culture, economy and set of institutions. What works in Korea may not be what Denmark should build, even if Korean children score higher on a test.

Having concluded that ranking school systems is extremely difficult, I naturally decided to make a ranking anyway. I did not think a spreadsheet could settle the question, but the alternatives seemed worse. News stories tell us what has gone wrong this week, and personal experience is real but narrow. If I wanted a broader picture, I had to look at the data.

There is something of Hans Rosling in that instinct. Our picture of the world is easily shaped by what makes the news, and what makes the news is usually unusual, dramatic or problematic. Data do not give us a final, objective truth either, particularly in an area as culturally dependent as education, but they can help us get closer to a reasonable description of reality.

So I took the latest PISA data and combined them with a broader set of indicators. The result is a ranking, but I think of it mainly as an analytical tool. Its value lies less in determining whether Denmark is eighth or eleventh than in showing patterns and pointing to places where it is worth looking more closely.

Once I started doing that, another question kept appearing behind the numbers: “How much childhood are we willing to spend in order to produce better educational outcomes?”

PISA is not really the problem

One of the first things I discovered was that PISA itself is more sophisticated than the public debate around it. PISA 2025 tests 15-year-olds in science, mathematics and reading, but it also collects data on inequality, belonging, bullying, attendance, teacher support, classroom discipline and school conditions. The OECD is therefore not claiming that three test scores tell us everything there is to know about education. That reduction tends to happen afterwards, in the way the results are discussed.

It is easy to see why. A league table is simple: one country has 510 points and another has 490, so the first is doing better. Once you look at more than test scores, the countries stop lining up so neatly. A country can achieve excellent academic results while putting students under considerable pressure, another can spend large amounts of money and produce fairly ordinary results, and a third can be unusually equal without being particularly strong academically.

At that point the main difficulty is no longer finding data but deciding what a school system should actually be good at.

What should count?

Some of it seems uncontroversial. Children should learn, and a school system in which pupils are happy but leave unable to read properly or understand basic mathematics would be hard to defend. Academic performance therefore has to count for quite a lot.

Other things clearly matter as well. How strongly does a child’s family background determine the outcome? Are pupils bullied, do they feel that they belong, and do they attend school regularly? These measures are sometimes treated as softer than test scores, but they say a good deal about whether the institution itself is working.

Then there is the classroom. A national curriculum and good intentions do not help much if there are not enough teachers, or if the teacher spends half the lesson trying to establish enough order to teach. Teacher shortages, the support pupils experience from their teachers and the disciplinary climate therefore also belong in the picture.

Parents matter too, although this is harder to score. Active parents can make schools work better, but a system that relies heavily on parental involvement may also amplify inequality, because parents do not have equal amounts of time, education, confidence or knowledge of how institutions work.

Money is another obvious factor. Schools need teachers, buildings, materials, special-needs support and everything else that makes an institution function. But spending is an input rather than an outcome, and the country that spends the most cannot simply be declared the winner. Luxembourg is a good example of this, and I will return to it later.

There is also a dimension that the current model does not capture well enough, which is time. More school, more homework, more tutoring and more examinations may produce better academic outcomes, but at some point the educational question becomes a cultural one: how much of a child’s time should be organised around performance, and how much should remain outside it?

This is where the question about childhood comes in. It is not an argument for asking less of children, since a school system can also fail children by expecting too little of them. It is simply a recognition that educational achievement has costs, and that not all of them are measured in money.

Building the comparison

I used PISA 2025 and Education at a Glance 2026 to compare 39 countries, of which 36 receive an overall rank. Costa Rica, Slovakia and the United States are excluded because too much of the underlying data is missing.

Academic performance has the largest single weight, but the calculation also includes socioeconomic equality, belonging, bullying, truancy, teacher support, classroom discipline, teacher shortages, student-teacher ratios, expenditure and material conditions. I have not included every variable I looked at. Class size, for example, is useful context but too crude to treat as a straightforward measure of quality, and internationally comparable measures of workload and parent-school involvement are not yet strong enough for me to include them in the main calculation.

The model has no special name; it is simply a way of combining the available data. The weights add up to 90, and the resulting country score is reported on a scale from 0 to 100.

Here is the current top 15:

Rank	Country	Score (0–100)
1	Korea	81.6
2	Japan	77.4
3	Finland	76.0
4	Canada*	69.0
5	Iceland	68.9
6	Switzerland*	66.7
7	Sweden	66.7
8	Denmark	63.7
9	Austria	63.4
10	Ireland*	62.6
11	United Kingdom	61.7
12	Türkiye	61.4
13	Norway	61.1
14	Estonia	59.9
15	Spain	59.4

* Canada, Switzerland and Ireland are missing part of the underlying data, so their positions should be read with more caution.

The decimal points suggest more precision than the data can support. There is no meaningful sense in which Denmark’s 63.7 shows that its school system is better than Austria’s at 63.4. The table is useful mainly because it sends us back into the data to understand why countries have ended up where they have, and that is where I ended up spending most of my time.

Korea and Japan

I had expected Korea and Japan to perform very well academically, but I had also assumed that both would fall once the comparison became broader. They did not: Korea comes first and Japan second.

Korea combines very strong academic results with low truancy, strong teacher support and orderly classrooms. Japan is harder to explain. It has the largest classes in the comparison, at around 32 students, and it ranks last of all 39 countries on both reported teacher shortages and material shortages. Only 3.5 per cent of Japanese students attend a school whose principal reports no teacher shortage.

Japan is nevertheless second overall. Conditions appear stretched on several measures, while the classroom itself seems to work very well: students attend, teachers appear able to teach and academic results are excellent. Counting teachers and resources clearly does not explain everything that happens inside a Japanese school.

What the data do not cover is what happens outside school. Both Japan and Korea have demanding examination cultures, and Korea in particular has a very large private tutoring sector. With a good internationally comparable measure of the total time children spend studying, the ranking might look different.

Still, the data make the usual Western explanation look a little too convenient. It is easy to describe East Asian educational success as the product of discipline, pressure and exhausted children, and there may be some truth in that. But it does not explain why the schools themselves appear to function so well. If the model has a cost, we need to find out where that cost actually falls rather than assume we already know.

Finland

Finland comes third, and the result is almost boring in a way I mean as a compliment. It does not dominate any single part of the comparison. Academic performance is strong, equality is strong, and classroom conditions and resources are good, with no part of the system lagging badly behind the others.

For years Finland was treated as an educational miracle. When its PISA performance declined, the miracle was widely declared over, which is roughly how the cycle tends to work once international rankings become news stories. Seen through a broader set of indicators, the Finnish story is less dramatic. Finland appears to have built a school system that does reasonably well at many things at the same time, which may be more valuable than being exceptional at one of them.

Denmark

Denmark comes eighth. As a Dane, I find this the part of the comparison that is hardest to reconcile with the way we tend to describe our own schools.

Academically, Denmark ranks 22nd out of 39, just below the middle. On belonging it ranks 17th, and on socioeconomic equality 14th out of 38. These are respectable positions, but they do not support the idea that Denmark has found some uniquely enlightened educational formula. Bullying is more uncomfortable, with Denmark ranking 33rd out of 37, and truancy is a little above average.

The most striking contrast concerns teachers. Denmark reports fewer teacher shortages than any other country in the comparison: 61 per cent of Danish students attend a school whose principal says there is no shortage of teachers, compared with 13 per cent in France. Yet Denmark ranks only 26th on teacher support and 28th on classroom discipline. We seem to be unusually good at getting teachers into schools without being particularly good at creating classrooms in which pupils experience strong support and good working conditions.

That seems worth more attention than another discussion about whether Denmark has moved three places in PISA, although experience suggests we will have that discussion anyway. It also shows why the ranking is most useful as the beginning of an analysis. Denmark’s eighth place does not tell us very much, whereas the combination of first place on teacher availability and below-average results on teacher support and classroom climate tells us considerably more.

The comparison also complicates the conversation that started this whole exercise. The UK, which is the closest national dataset available to the English schools I had heard about, comes 11th overall, only three places behind Denmark. It performs much better academically and also does well on the equity measure used here, while Denmark performs better on belonging and has far fewer reported teacher shortages. PISA covers the United Kingdom rather than England alone, so the comparison is imperfect, but it is a useful reminder that one family’s experience and a national school system are not the same thing.

Estonia

Estonia comes 14th. If the question is mainly about academic performance, it is one of Europe’s most impressive education stories, but it falls once the definition becomes broader.

Its equality score is around the middle, at 15th out of 38. It is second worst on bullying, at 36th out of 37, and second worst on reported teacher shortages. It also spends less per student than most countries in the comparison, and its schools report more material shortages than most.

None of this makes Estonia’s reputation undeserved, but it makes the achievement more specific. Estonia is very good at producing strong academic outcomes with comparatively limited resources and in fairly difficult conditions. That is a different achievement from having the best school system overall, and possibly a more useful one for other countries to study.

Türkiye

Türkiye is one of the less expected countries in the upper part of the table, where it comes 12th. A large part of the explanation is teacher shortages: Turkish school principals report very little of it, which puts the country second out of 39 on that measure. At the same time, Türkiye ranks 34th out of 36 on truancy and has the second-lowest expenditure per student in the comparison.

Teacher shortages carry substantial weight in the calculation, but the measure is based on school principals’ own responses. Japanese principals report more shortage than anyone else, while Turkish principals report almost none. Some of that may reflect a genuine difference, but some of it may reflect different expectations about what counts as a shortage, or about when a school leader regards a staffing situation as a problem. Data do not become culturally neutral simply because they end up in a spreadsheet. To understand the difference properly, you would probably need to talk to school leaders in both countries rather than rely on their survey answers.

France

France comes 30th out of the 36 countries that receive an overall rank.

Only 13 per cent of French students attend a school whose principal reports no teacher shortage. France also ranks 35th out of 39 on classroom discipline and 29th out of 38 on teacher support. This fits at least part of what French students have been complaining about in the current protests, where teacher shortages and deteriorating learning conditions have been recurring themes.

Family background also matters considerably. The difference in science performance between socioeconomically advantaged and disadvantaged students is about 100 PISA points in France, compared with about 69 in Denmark.

None of this proves that every complaint made by a protesting French student is correct, any more than the experience of one Danish family in England shows that English schools are bad. But when the anecdotal observations are set against the broader data, they look less like isolated stories about individual schools, because some of the problems are visible at the level of the system. I started this exercise with people complaining about schools and went looking for something less anecdotal, and in France at least the data brought me back fairly close to where I began.

What money can buy

Luxembourg shows the limits of spending more clearly than any other country in the data. On the expenditure measure used here, it spends USD 32,766 per student, more than any other country in the comparison, yet it sits close to the bottom of the overall ranking. Japan is a useful comparison in the opposite direction, since its measured resource conditions look unusually strained and it still ranks second.

It would be wrong to conclude from this that money does not matter. A school without enough teachers, decent buildings or basic materials obviously has a problem, and below a certain level additional resources are likely to make a large difference. But resources are not the same thing as a functioning school system. Once the basic conditions are in place, what matters more is how the school is organised, what is expected of pupils, the authority and competence of teachers, and the relationships and culture inside the classroom. Money can pay for the inputs that make good education possible, but it cannot guarantee the outcome, as Luxembourg demonstrates at some expense.

The problem with my own ranking

The obvious objection to the table above is that I chose the weights. I gave academic performance more importance than expenditure, teacher shortages more importance than student-teacher ratios, and so on, and someone else could reasonably make different choices and arrive at a different order.

To see how much this matters, I tested alternative versions that gave more weight to academic performance, to the child’s experience of school, or to teachers, classroom conditions and resources. The results changed, but not equally for every country. Korea, Finland and Japan stay close to the top across the different versions, and Denmark moves within a fairly limited range, while Iceland, Norway and Estonia move considerably depending on what is emphasised.

This is useful to know. It shows which countries look strong under several definitions of a good school system and which are very good at particular things, and it makes clear why a precise rank should not be confused with a finding about objective educational quality. A ranking cannot remove the values that go into it, but it can make them visible. For me, that is the best argument for making one, because it forces us to state the question more carefully and gives us a way of finding countries whose combinations of results deserve a closer look.

School as a portrait of a country

At this point the exercise becomes less about educational statistics and more about culture.

School systems are one of the places where societies turn abstract values into everyday practice. Ideas about equality, authority, competition, responsibility and the role of the state end up as timetables, classrooms, examinations, teacher roles and expectations of parents.

This is one reason it is difficult to separate a school system from the country around it. Danish schools belong to the same society as the Danish welfare state and labour market. Japanese schools belong to a society with a different relationship to hierarchy, collective responsibility and academic effort, and French schools grew out of yet another political and educational tradition.

Looking at schools can therefore tell us things about countries that GDP figures or opinion polls cannot. They reveal what adults believe children need to become, how much authority teachers are given, how responsibility is divided between families and the state, how much inequality a society is prepared to tolerate and how hard children should be pushed in order to improve their chances later. These are educational questions, but they are also questions about what kind of society we think we are building.

How much childhood?

The trade-off is most visible when we think about time.

A school can ask too much of children. Long days, homework, tutoring and constant competition can take up a remarkable amount of childhood. But the opposite position is not automatically the more humane one. A system that asks too little may give children more freedom at 14 and leave them with fewer opportunities at 24.

Childhood has a value of its own, but it is also a preparation for what comes after it, and a school system has to take both seriously. Children need time that is not organised around measurable educational outcomes, for friendships, sport, interests that lead nowhere in particular, wandering around and occasionally doing very little. They also need to learn difficult things, practise things they would rather not practise and discover that effort sometimes matters more than immediate enjoyment.

There is no formula that tells us where the right balance lies. Different cultures have arrived at different answers, and families within the same culture will disagree about it. This is why the question I started with became harder rather than easier the longer I looked at the data: “How much childhood are we willing to spend in order to produce better educational outcomes?”

It is tempting to expect an international ranking to answer a question like that. It cannot, because part of what we are comparing is the different answers societies have already given to it.

The French students blocking their schools were not protesting about a PISA table. They were asking whether their school worked: whether the teachers showed up, whether teaching happened, whether conditions were reasonable and whether the time they spent there was worthwhile. That seems to me a good starting point for deciding whether a school system works, even if it is much harder to rank.

Method and data

The comparison uses data from PISA 2025 and Education at a Glance 2026. It covers 39 countries, of which 36 receive an overall rank. Costa Rica, Slovakia and the United States are excluded because too much underlying data is missing.

Each indicator is scaled from 0 to 100 within the comparison group, with extreme values capped at the 5th and 95th percentiles. Expenditure per student is log-transformed before scaling to account for diminishing returns.

The indicators are combined using the following weights:

Measure	Weight
Academic performance	20
Avoiding severe low performance	5
Socioeconomic equity	15
Sense of belonging	5
Bullying	5
Truancy	5
Teacher support	7
Disciplinary climate	8
Teacher shortages	10
Student-teacher ratio	2
Expenditure per student	2
Material adequacy	6
Total	90

The overall country score is reported on a 0–100 scale. Canada, Switzerland and Ireland are marked with an asterisk because part of their underlying data is incomplete.

Teacher shortages and material shortages are based on school principals’ own responses. They are therefore subjective measures and may partly reflect differences in national expectations and reporting practices.

The same caution applies to student-reported measures such as belonging, teacher support and disciplinary climate, where national expectations and response patterns may also affect cross-country comparisons.

I also tested alternative weighting schemes that placed more emphasis on academic outcomes, the child’s experience of school and the conditions around teaching. The purpose was not to identify another correct ranking, but to see how sensitive the results were to different definitions of quality.

The data collection, cleaning, calculations and parts of the comparative analysis were carried out with the assistance of artificial intelligence, which made it possible to collect and test a much larger international dataset in a relatively short time. The source data come from the OECD; the selection of indicators, the weighting, the interpretation and the conclusions are mine.

The underlying data and the country scores are available in two accompanying files: raw data and country scores.