I think there's a general presumption that things can be improved, without a great deal of evidence of that. Second, government often delivers through social service agencies anyway, and often very efficiently, but there's no baseline evidence of that. The general assumption and presumption behind social impact bonds is that there is incompetence and inefficiencies that can be improved, and this is the way to do it. I would challenge that.
On March 26th, 2015. See this statement in context.