One of the the bonds' main pluses is that it doesn't specify how a particular policy objective shall be achieved. So when the relationship between cause and effect is obscure, and when there are a multiplicity of possible solutions to a problem, a Social Policy Bond approach might be preferable. As society grows ever more complex, I expect such policy areas to increase in number. I have argued at length about the need for outcome-based policy to deal with climate change. I also think that something like Social Policy Bonds could be issued to eliminate violent political conflict, whose causes vary wildly and are extremely difficult to anticipate and close down.
Less grandly, I am also interested in literacy in the developing countries. In Bangladesh something like 20000 non-governmental organisations operate, many of them supplementing the Bangladeshi Government's literacy initiatives with their own. With literacy-raising, most methods will be well-tried and tested, but at a country level there is scope for diversity and experimentation. Social Policy Bonds, unlike centrally directed projects, would encourage such adaptive approaches because of their inextricable linking of rewards to efficiency. They could lead to a mosaic of different approaches over a whole country, each precisely targeting the local circumstances in which they operate. Something like that his happening now in Bangladesh, but unfortunately (as I see it) with little orientation toward a national objective. Thus, many of the non-governmental organisations, principally those financed by Islamic institutions, favour the teaching of boys over girls, and literacy in Urdu (which is the language of Pakistan), rather than Bengali.
A Social Policy Bond regime would have a single, unifying, literacy objective, and this would be quite compatible with diversity of approach - if such diversity were found the most efficient way of achieving it.