This concept is similar to the idea of social impact bonds (SIB) / pay-for-success (PFS) contracts / flexible prizes to incentivise impact investors to fund the development of public goods in return for the outcome payments if successful. Crowd Funded Cures is working on implementing this idea to incentivise funding of clinical trials for “unmonopolisable therapies” where patents do not work because you cannot recover your investment by charging a monopoly price e.g. finding new uses for off-patent drugs, supplements, plant medicines & psychedelics, diets and non-drug / lifestyle interventions—see https://crowdfundedcures.medium.com/pay-for-success-contracts-a-new-model-to-develop-new-therapies-from-old-drugs-f69b2189184d
This creates a scalable business model on the basis that the cost savings will exceed the amount of outcome payments. For example, a SIB for homelessness, a local govt pays $10k per homeless person housed for at least a year, which saves the local govt $100k in reduced policing costs and economic harm. In the same way a $50m SIB for obtaining FDA approval for repurposing off-patent fluvoxamine to treat Covid could save governments $1b+ from reduced reliance on patented drugs such as molnupiravir—see https://golab.bsg.ox.ac.uk/community/blogs/innovative-financing-mechanisms-2/
A flexible prize model is a much simpler design for creating a market for doing public good / charitable things. Seems that terminology being used is complicating matters—whether you call it Impact Certificates, Social Impact Bonds, Pay-For-Success contracts, Advance Market Commitments, Retroactive Public Goods Funding , etc. They are all flavours of conditional contracts . But you want to ensure that the parties cannot change the terms of the deal for receiving outcome payments unilaterally in the future (which IMO is a problem with RPGF).
The part which is the most important is for a funder to put a price on some measurable outcome that they want and have clear and unambiguous criteria for determining whether the outcome is met (e.g. $10k per homeless person housed for at least 1 year). Ideally, this outcome should provide more benefit to the funder than the price paid (e.g. $15k+ cost savings due to reduced policing, boost to economy etc), which makes it an arms-length negotiation and scalable business model. The superintelligence of the market does the rest (e.g. VC investing in companies competing for the payments, and automatically price in what they think are their future cashflows etc). Where it can get more subtle is where you are paying a flexible outcome payment proportional to impact.
The other concern is that if we assume funders will have a limited budget, so need to prevent blow outs. In that case, you can pay a fixed amount distributed between companies “registered” to the fund that achieve the criteria for outcome payments, proportional to the achievement of their impact measure (e.g. each person put in a home provides a number of “points” that determines the company’s percentage entitlement for a pre-determined period of time such as 5 years). Market equilibrium is reached with a certain number of company registrations, and the proportion of rewards available proportionally increases as the “points” cease to be eligible for outcome payments after the pre-determined period of time . Social impact is then infinitely scalable with the size of the fund, and the market automatically allocates payments in an efficient way.
So to illustrate this in an example:
A funder decides they will pay $2m p/a in a flexible prize fund, allocated between companies registered to the prize fund proportional to the number of homeless people housed for at least 1 year. Company A houses 30 homeless people for a year (30 points). Company B houses 90 (90 points). For that year, the £2m will be split 25% to Company A and 75% to Company B. They will keep those points for 5 years. Next year, Company A houses another 30 (30+30 = 60 points), and Company B houses another 90 (90 + 90 = 180 points). Same split. In the third year, Companies A and B they do the same, but a newcomer Company C puts in 100 people in a home, so it will be split 90/270/100 between A, B and C. After 5 years, Companies A and B lose 30 points and 90 points respectively as the 5 years is up, which provides more proportionally for everyone else.
If a funder wants to increase the number of homeless people housed, they just increase the annual payment, and a new “market equilibrium” is reached.
This flexible prize model can be applied to any form of public good, as long as we can agree they are robustly measurable and comparable with each other (e.g. QALYs, % reduction in hospitalisations vs usual care, number of kids put in college, lives saved etc).
Credit to Pogge and Hollis for the idea of a flexible prize fund: https://en.wikipedia.org/wiki/Health_Impact_Fund