But I happen to think that formal education beyond the primary school level is mostly a sorting mechanism with elaborate signalling races developing around it.

This flies in the face of my experience. And I do believe I can cite evidence against it but it is not from studies per se.

I actually had a very good public school education, in a well off suburban school district in 1960s and 1970s Farmingdale, New York. In addition to this being a well funded district that people moved to for its reputation, I was sorted from 3rd grade through 9th grade into a special program with even better teachers and curriculum aimed at the top 2% in IQ terms of the district.

Even with that excellent background, it was not until Swarthmore College that I understood why missing a day of class could possibly be of any concern (because the pace was so fast). Further, I would estimate that I learned, hard to say, 10X as much science and math in 4 years at Swarthmore than I had in 12 years of Farmingdale.

I then spent 2 years as a technician working with radio astronomers at Bell Labs. Yes, I learned so much there that for a while I would tell people I did my undergrad at Bell Labs. But to be fair, it would probably be more analagous to having done my MS at Bell Labs. My undergraduate education was superb.

I then went to graduate school at Caltech, and surprisingly instead of signalling my sorting, I learned and did independent research, presented my results to the elite, and developed into somebody who could reasonably say that if he didn't understand a particular physics/astronomy presentation, that it was the presentation that sucked, not me. Yes, I learned LOTS MORE even in graduate school, in math methods and statistics/probability, and quantum mechanics and classical mechanics and, holy of holies, electromagnetics, my own specialization. Oh and in solid state physics, I was working on superconducting devices.

So my own experience is I learned gobs and gobs and GOBS of stuff in college and gradual school.

Does the possibility exist I could have learned this on the job? Sort of but not really. First off, classes were INTENSE, much harder than job. Second, jobs you tend to learn what you need to do your immediate task, classes front-load you with stuff, some of which you may never use, but lots of which you wind up trotting out over the next 50 years of your career.

So that's my personal experience. What is the evidence beyond that that school is real, not just signalling?

1) Greatest researchers tend to be at universities. They tend to prefer to work with students and post-docs (still an educational job IMHO).
2) Even the most iconoclastic geniuses in math and science will typically skip lower levels of education, but be brought in at unusually young ages to the highest levels of education. Wolfram graduated with a PhD from Caltech when he was 21 (IIRC) and was hired as a professor by Caltech at that point. Physicists and Mathematicians seem to want people from these programs, and it doesn't seem to me primarily because they are trying to skip having to make their own decisions on who to work with.
3) I interview for engineering positions at my company. We have information on where applicant came from and what degree we have, but our interviews are still primarily technological tests. We want to know if applicant knows CDMA and LTE and 802.11xxx and GPS and AWGN and dB and Matlab and all the things we associate with people who get stuff done. As a participant with others in the interviewing of others, I have NEVER heard a discussion of the quality of school or which degree the applicant had. It is always a substantial discussion. And our ranks of hires include people with BS and years of experience, as well as MS and PhDs with less work experience. At some level X years of experience seems to equate on average to Y years of higher education. But the point is both experiences are different and both are valuable in employees. We have plenty of PhDs from great places, plenty of PhDs from podunk places and plenty of BSs.