[PeterisP]

IDK, the main part of the PhD is the work that goes into your research papers; if you leaver your current program but already have sufficient publications, in many universities there's a process (not commonly used, but a legitimate option still) where you just write them up in a thesis and go straight to a defence.

It would be hard to switch if you have a year's work invested in some research that's not published yet, in that case much of that effort "wouldn't count", so switching once you're a year in or so loses progress, but if you're 5/6ths ready, then that's different. Also, what's 5/6ths ready at your institution probably is completely sufficient at a bit less prestigious institution - not optimal, but it's important to know that there is a reasonable exit, a Plan B that allows you to leave abuse.

[mattkrause]

It's a bit circular: one of the major reasons people want to leave a group is because it is not productive. If your PI won't invest in data collection or sits on manuscripts for years, you won't have the raw materials for a thesis-by-publication.

As for transferring, other than a lab moving, I've never heard of someone switching just to defend. Most programs have "residency" requirements of a few years (ours is two years after a master's and at least 2/3s of the coursework, for example).