Social impact bonds are one of those excellent ideas that are great in theory but struggle in practice. This post will look at the challenge with SIBs as currently practiced, and the next one will explore how Web3’s ‘bounties’ could offer (some of) the answers.

Social impact bonds aim to channel the market to positive outcomes.

The idea behind social impact bonds (SIBs) is compelling - identify needed social outcomes, and have funders (generally governments) commit to pay the private sector (for profit or non-profit) innovators only when the outcomes are achieved. With a pledge for future payment in their hand (and a plausible theory of change - more on that later), these innovators can raise the up-front, operational capital from the other funders, such as the private sector banks looking to support impact projects, who get paid back when the outcomes are delivered.

The UK has been a leader in this area, championed by venture guru-turned social impact entrepreneur Sir Ronald Cohen, and launched the first effort by which non-profits were rewarded if their activities reduced prison recidivism.

Challenges with SIBs

There are two categories of challenges with SIBs, operational and philosophical, which has meant they’ve not been adopted at scale.

Operational challenges:

• Set up and legal costs. Legal costs tend to be significant - each one requires lots of paperwork and lawyering.

• Monitoring costs. Often the bonds require an expert, credentialed group such as a university to vouch for the theory of change and also oversee the implementation.

• Lack of metrics & common standards. Measuring results is hard - what is a unit of “recidivism” or “better health” outcomes?

• Long cycle times. If you’re paying providers to reduce recidivism you need a time frame. A one year window for reoffending or 30 day re-admission to hospital seems reasonable, but what if everyone starts re-offending or turning up at hospital a day after the deadline expires and the provider’s cash is in the bank? Or for a bond around healthy aging, you’d need to monitor it for years.

• Lack of contextual data. It’s often hard to get the data from the front line and many of the bonds focus on the digitally disconnected, who are not wearing Apple watches and posting daily on Instagram.

Philosophical challenges

• Outcomes obsession. Sure what gets measured gets managed, but we risk boiling things down to numbers and overlooking the human dimension. Someone living alone and in need of help may have 5 visits a day, but if these visits are 5 minutes box-checking exercises, it’ll be miserable for all involved.

• Risk-avoidance. If you’re just paying for outcomes, you really don’t need to worry about how they solution is implemented. Yet it seems most SIBs have used tried and trusted theories of change to de-risk their proposal to funders. What’s the point of a SIB if we already know what needs to be done?

• Role of government vs private sector. Should the government be outsourcing social programs to the private sector? Especially if they have a good idea of what works and what doesn’t (see Risk avoidance above).

• Delegating priorities to the private sector. SIB projects will be funded if the innovators and up-front capital providers prioritise these topics. The governments then are following the priorities of the private sector, which isn’t how social policy should be made.

In essence, social impact bonds are trying to apply market incentives and mechanisms to a slippery blob of normative issues and topics. The tools they’ve been using - bespoke, paper-based contracts struggle to turn the messy reality into something that is programmable.

Enter Web3. While Bitcoin was brilliant in introducing a scalable concept of digital money, Ethereum introduced smart contracts, and Chainlink introduced oracles that allow smart contracts to interact with the real world. All these are breakthroughs that could be relevant to SIBs; the next post will look at this, in particular the use of ‘bounties’ to address the fund outcomes; whether it’s tech coding or desired impact metrics.