If a school only accepts students capable of attaining high grades, in what sense is that school good? That's no proving the school teaches well, it's proof that they're trying to avoid having to teach.
Being selective means they can start further, go faster, and expect motivation.
Even if the value is coming entirely from your fellow students, that has no effect on deciding which school you want to go to. You want to be in that environment, and also get the status from the exclusivity.
I've studied in not-so-elite math/physics/CS high school in Russia. It has a focus on full coverage of school program so alumni would crush the exams to get into universities they want. The entry exams were hard because in the first year they cover the entire program of specialties without cutting the corners (i.e. van der Waals equation, nearly all known to science properties of triangles and 4-side polygons, methods to ease brute force a polynom of Nth degree) with extensive homework. This is an impossible task for anyone who isn't already good because there's so much to learn and only 24 hours in a day.
The actual quality of the education at these schools varies. I went to one of the ones mentioned, and had a mix of good and bad teachers (including an awful math teacher one year, and an awful informatics teacher in another year - two of the supposedly most important subjects for that HS).
However, having an entire class that actually is at the bar for which the curriculum is designed, and having an entire class of students who generally want to learn, does allow teachers to actually spend most of their time teaching the curriculum as intended. We had no problem being up to date with the Bacalaureat matter, for example. No one in the entire HS flunked their exam either.
The general state of high schools in the country is terrible though. Teachers don't care to know the curriculum, they don't have time to go through it if they want anyone in the class to actually be listening, many poor kids don't have time or energy for school among their many other chores (especially in the more rural areas of the country, children are a big part of house and field work) etc. Many children who nominally finish high school probably haven't even heard of a quarter of the questions they'd get on the Bacalaureat.
It's also important to note that there is a massive cheating culture in Romanian schools, all the way up and down. There was some push ~10 years ago to stop cheating at the national exams, installing cameras and such, and the Bacalaureat results dropped from ~80% pass rates to ~40% nation wide in that year; there was one case of an HS where the previous year 90+% of the students had gotten a 10/10 on their exam, and after the anti-cheating measures the following year, not one child got the 5/10 required to pass.
The Capacitate/National Tests exams that you take at 14 were even worse, with all sorts of advanced questions that were either cheated through or at the very best rotely memorized. In my time, we were required to do literary analysis on one of these exams, writing a 2 page essay on things like "the characterization of <female heroine in major novel>". After some reform, today you get questions like "here is a newspaper article about a meteor. How large is the meteor? What is it made of?" and still many many children don't pass - because those are the actual results when you take cheating out.
Not to mention, Romania consistently ranks last in the EU at PISA tests. And around 50%+ of the country is functionally illiterate (that is, they technically know the letters and words, but can't actually read an article and tell you what it means).
Your conclusion assumes the PISA test actually reflects students' abilities. This may not be the case. My daughter when attending a Romanian school told me that students know the PISA results play no part in their performance rating and many of her colleagues treated the exam either as a joke maybe even supplying deliberately dumb answers or at best as an irrelevance - 'shove anything down and finish with it'. Her experience was that the teachers were, let us say, not helpful, in countering this attitude. So, incorporate the PISA results into students' records and see if the test results change.
A number countries have stats that account for the socio-economic background of the kids. In the UK it's called progress-8. It looks how how good a school is vs what would be expected from the mix of students they have.
