One of the benefits of social impact bonds is that they can create a discipline, as you say, around the expected outcome. You're exactly right.
Today, we can track every cow from birth to burger in Alberta but we can't tell you how many people experience homelessness in Canada, what happens to them when they're in the system, and what happens to them when they leave. We can do that in every other field; there is no reason why we can't do it for homelessness. We've begun to use systems here in Alberta called homeless management information systems. We should be able to track everybody. We know who they are and what they need. We have to move them into the system, track what happens to them, and understand what happens to them when they leave the system. But in social services, that data and those data systems are very poor.
A good example of a decent system that's not functioning as best as it could is the federal homeless individuals and families information system. That system collects data on people who are in federally funded programs, but it's very difficult on the ground, in agencies, to have knowledge of who those people are and how they move through the system. Information goes up but it doesn't flow across, so we can't actually track people through the system of care.
Until you're able to track that data and know for a fact.... For example, to prove success on a social impact bond for homelessness, you have to prove that a person is housed and has stayed housed for a year, and define how their use of the public system has been reduced. But today, without a homeless management information system, we can't track whether or not that person falls back into homelessness. In my view, you have to have a system-wide coordinated data system.