Your parent does not really say that though. Assuming that WFH/RTO preference is distributed equally across highest and lowest performers his statement will be true. As lowest performers will be stickier (regardless of preference) and highest performers won't be so you may lose 50% of your highest performers but a lower amount of lowest WFH-loving performers. Effectively averaging down the skill level of your company.
It's not really a WFH/RTO thing so much as a benefits thing. The same would be true for another variable like free lunches.
A study like that would be nearly impossible to measure.
What we do know is anecdotal evidence of very influential and high performers at various FAANG type companies that have ended up leaving due to forced RTO.
Everyone has anecdotes that support their opinion.
Anecdotally, I work at a large company that fully supports 0% to 100% remote work, and I work with a wide mix. Some people, and some roles, just do better in the office. Others don’t. I’d be very suspicious of forming beliefs based on anecdotes, especially if they coincidentally align to your expectations.
I do agree some roles. e.g. sales, have to be done in office. If you want to snooze someone over dinner a remote call is a poor shadow of an option.
Less certain about roles that are generally technical. The strongest claim for in-office work, collaboration, is mostly a well solved problem for at least software, possible simple administration that just relies on some collaborative spreadsheets. Sitting everyone in a stuffy boardroom all staring at the same boring spreadsheet is likely to cause stress and mental fatigue, not increased collaboration. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7889069/ This is a well known effect.
Even large "interactive" meetings in remote settings can be counter productive - of course not all meetings are bad and one key to being an excellent manager is picking the right people for the right meetings, but the idea that your "top performers" objectively speaking will be the ones experiencing the most stress is seriously suspect.
What is more likely is some combination of workers who either "do well" in high stress environments, usually implying they are healthy and have some effective stress management techniques and as such are recognized as top performers compared to those unable to effectively manage their stress, or those in mainly technical areas have not yet experienced (somehow) the benefits of less stressful remote-work environments. E.g. parents or partners in cramped personal quarters being more distracted compared to e.g. a private office space.
Finally, for those workers that have an active job component where they must e.g. be on site for customer testing or perhaps they are a technical liason, some hybrid approach may work better - in office for resources which are cost prohibitive to simulate or when their role requires them to interact in an another capacity e.g. as a technician, security guard, or triage such as a front desk.
There is also the case where "top performers" are leeching off their colleagues e.g. where they can "get more done" because they can interrogate and corral others but where others are mostly experiencing productivity hits as a result of the constant interruption.
It's not really a WFH/RTO thing so much as a benefits thing. The same would be true for another variable like free lunches.