As an example, I'm currently working on a proof of concept using Rust to replace some existing C++ that coordinates some Gstreamer pipelines. This sounds like the perfect task for remote: self contained task, redoing existing code, fairly small scope. But remote (or more specifically the time difference) makes it pretty difficult. I don't know Gstreamer that well, and I don't know the existing code that well. I'll often run into a problem that I KNOW someone on my team could answer in 2minutes, and unblock me. Instead I need to push through wasting valuable time trying to figure it out myself, or wait till they are online. Perhaps this is the time difference not the remote aspect at fault. If they were on the same timezone, it'd be a quick Slack message away
Now, I am reduced to writing code that slots in around my team. As for actual empirical metrics, I don't have many. We tend to track achievements as a team (OKR's etc), and my team has been hitting there goals mostly. I do feel we could achieve more though if I was onsite. A probably non valuable one might be Github commit count, which dropped from ~ 3500 for the 2018 year to ~ 300 for the last 6 mo. So at least on code alone, I am churning substantially less of it out.
I'm not poo-pooing remote btw. I think my story would be totally different in a 100% remote company, or a company with substantial remote presence. I think for me though, in this situation, I prefer on site.
Not to say that your issue might not be remote in general, but for me I definitely think that having much more than the 3 hour difference I currently have with my team would be tricky: just too much time where someone is waiting for someone else.