Yes. I think part of why governments are interested in SIBs in the first place is this idea, not of getting something for nothing, but that they should only pay for outcomes in the end.
I think the research out there that we've compiled and the literature that exists now show that there are short-term costs associated with creating the infrastructure to develop SIBs, including new skills for bureaucrats and additional costs for lawyers and accountants, as well as evaluation. I think John talked to how important that evaluation component is, because that's what triggers the repayment. Those are all costs that if not retained in-house would need to be contracted out. That's something that's often not talked about.
Also, of course, there are the long-term costs. Government will have to pay over the long term for these outcomes. There are actually two different types of SIB models. One is a frequency scheme, whereby investors would receive a repayment every year based on the results of that year. One is a binary scheme, where investors receive repayment at the end. There's a question as to when government would do that repayment process. Would it be at the end of the project or would it be over the course of the project? Anyway, government will have to pay for those results as well, and then the question is, would that be at higher private sector interest rates than government would normally pay for its own financing?
These are other elements that we need to thing about in terms of the costing.