How efficient is net neutrality?

Net neutrality, the fact tat no one has priority over the use of the Internet, may sound very democratic, but it is also potentially inefficient. A Skype communication, which cannot afford much latency, has the same priority as an email, which can definitely afford to wait a few seconds. The obvious solution is market based: let people pay if they want less latency. This should also make the broadband hogs who need to watch videos absolutely everywhere realize how they are affecting the Internet use of others. But that would give up the ideal that the Internet is free for all and whichever way you use it.

Nicolas Curien points out that net neutrality does not imply that Internet use should remain free, both for end users and content providers. It only implies that the price does not discriminate in any way. Curien goes on to formalize in a rather convoluted way neutrality and efficiency, showing that the most efficient outcome in many situations implies some imperfection in neutrality. This is hardly surprising. Maybe more interesting is the discussion in the conclusion, unfortunately not backed up by formal analytics, that competition makes neutrality easier to enforce, and thus Europe is in a better position in this respect than the United States. I am curious whether is true.