I'm asking this in a general sense; does it look bad for the supervisor if the PG student, especially the PhD student drops out from the programme?

Dropping out reasons I can think of at the moment are:
1)No funds
2)Poor supervision/No direction
3)Personal issues
4)Supervision conflict

You can add in any reason if you like.

Not really particularly when it's early on in the process. Unfortunately if you look at the statistics self-funded students especially are much more likely to drop out; the pressures in juggling so much are often much greater than the students expected at the beginning.And frankly, money pressures make any personal problems seem far worse. The students that supervisors do get grief over are those who stay in the programme for years but never manage to submit a PhD, or funded students who don't complete, because there can be implications with the funders for future students. Otherwise it's generally accepted that some people will realise that it's just not what they expected (particularly in the first 18 months) and leave. A PhD is so different from a taught degree that it's not surprising that you can do very well on the latter, but find the very independent working on a PhD hard to deal with for humanities/ social sciences types, or for scientists struggle to adjust to working in a team in a lab. It is different and people understand that. I think as well that for students who've gone straight through university to start a PhD, essentially it's like a first graduate job and many people struggle to adjust to that reality too, so there are always going to be those who have made the wrong choice career wise and want to jump.
So if this is something you're thinking about doing (as your last few posts have sort of suggested) don't let worries about it looking bad for your supervisor effect your decision, if the PhD is something that's not right for you at the moment.

Hmm...

People in my department seem to be proud that my PhD programme has an "attrition rate" — this is what they refer to it as, they seem too hard-core to refer to it merely as drop-out rate — of 50%. Actually, with the REF coming up, I think it might even be in some PhD supervisors interests for some PhD students to drop-out.

Yes. The supervisor may look bad for poor supervision. But PhD student does not look good too; a quitter. This PhD student did not choose supervisor carefully!

2) Poor supervision/No direction is not necessarily bad. PhD student should be independent!

3) Personal issues are common everywhere. We are all "wired" differently.

4) Supervision conflict: If the supervisor is very brilliant, then the student could be perceived as incapable and wasting the professor's time. If the student is more brilliant, then the supervisor could be perceived as useless. Moreover, supervisor and student may not be compatible in terms of thinking style. For example, men from Mars, and women from Venus.

My supervisor is quite weak. Luckily, my external examiners have good comments on my thesis. However, this thesis could have reached greater heights.

One suggestion is to ask other PhD students within the university about your supervisor. Just to make sure you are not biased.