Measuring proficiency is always a difficult subject.  The italki system for reporting familiarity with languages is certainly highly subjective, but I have always assumed that people downplay their stated proficiencies.  

I suppose there have been instances in which I have doubted a person's self-assessment, particularly when they are offering to teach other people the language.  As far as language exchange partners and general users are concerned, I do not feel that there is any real harm in these inconsistencies.  After all, what would be the alternative?

Well, I think we can agree that in spite of how members rate their own language skills, the proof is in the pudding.

I too have seen members regularly overrate (and underrate) themselves, and there are probably a couple of reasons for this. For example, maybe there is a misunderstanding of what the CEFR signifies, and I guess these people haven't even bothered to look at what C1 or C2 actually means. So, they make an arbitrary guess. On the other hand, I've seen members overrate themselves based on a good test score - which is usually from a multiple-choice test, which makes your language skills higher then they seem anyway, in my opinion.

Perhaps some rate themselves up because they think it makes them look good.  I've even seen people on italki who not only lie outright about their ability, but also their nationality.  They're easy to spot not only form their obvious non-native errors, but also their insistence that what they write is perfectly correct and that they are a "native speaker".

In the end, I don't give these people much heed. I'm not a policeman, and they can wallow in their self-delusion. It's only if they are being antisocial and sabotaging other people's progress (eg. by giving bad advice) that I'd react.

I've noticed it a time or two.  I try to do the opposite.  I'm still in the first few lessons for all of my languages, so I don't even use the ranking system yet to avoid annoying people in the manner you describe.