While crawling the Internet, I found a study asserting that PC norms boosted creativity in mixed-sex groups. The linked article talks about it in context of Jonathan Chait's article "Not A Very PC Thing to Say." According to the study, PC norms reduced uncertainty in mixed-sex groups by making the rules of engagement very clear. Since the rules - and the penalties for breaking them - were clear, the people in the groups could share more creative ideas without fear of looking bad.

My own belief is that the study is limited in that it only rates ideas for novelty and not feasibility, that the groups don't actually have to implement the idea, and the fact that the PC norm is imposed on all of society at all times. Also, the PC norm stifles discussion of topics relating to social conditions in various groups, be they races, sexes, religions, etc. Indeed, the topic discussed in the study was "what kind of restaurant to open" - something noncontroversial and safe.

The linked article tries to justify PC norms in society at large:

But I would argue that the Cornell study has a place in this conversation. It's measuring people's abilities to communicate with each other in productive ways, as well as the ways imposing constraints on speech helps or hurts that effort. It's not that big a jump to apply implications for workplace performance to the realm of political effectiveness.

Like I said, it's easy when you're discussing a specific noncontroversial task in a controlled environment (and you don't have to actually do the task yourself.) It's much harder when politically motivated norms are imposed on all in an effort to hide uncomfortable truths.