I'm sure the actual numbers in here are kind of crap, but I'll selection-bias with them because the qualitative results match my observations:

http://spin.atomicobject.com/2012/01/11/small-teams-are-dramatically-more-efficient-than-large-teams/

It talks about eg. a 4-person team getting things done in only slightly more time (and thus much less cost) than a 30-person team. Of course, it has to be the right 4-person team.