Social Impact Bonds: complexity wrapped in conformity
Last week I attended the Social Outcomes Conference hosted by the Government Outcomes Lab at the Blavatnik School of Government. It was a fascinating event, that aimed to provide “an in-depth exploration of the practice and evidence base around the implementation of outcomes-based models of public service provision from across the world”.
I was speaking on a panel which asked whether we’ve become “overly fixated” on social impact bonds (SIBs). Given that 21 of the 23 sessions over the two-day event were explicitly about SIBs it would appear the answer is a definitive “yes”.
I’ve long been interested in SIBs, and have written about some of their internal contradictions before, but have not been close to the SIB community in recent years. Returning to the debate again last week I was surprised to see how many of the old questions around risk transfer, outcomes measurement and pricing remain largely unresolved nearly a decade after the first SIB was developed. But perhaps that’s to be expected given the technical complexities of SIBs — something we’ve known all along.
What has changed over the past 10 years is the way policymakers think about outcomes. Previously, taking the cue from New Public Management, outcomes were generally viewed as part of a neat linear sequence with inputs at one end and impact at the other. Line everything up, turn the handle, and you’ll produce outcomes like sausages from a sausage machine.
This handy graphic, courtesy of the Government Outcomes Lab, sums it up.
But now many people are challenging this deterministic logic. Isn’t it the case, they argue, that this is a very poor representation of the messy reality of the real world? Rather than imagining that we can measure and control everything we should instead view the world as a complex system.
Using this framing, outcomes are emergent properties of that complex system and trying to “manage” or “deliver” those outcomes makes as much sense as managing the patterns of birds in flight.
At this point, many SIB proponents are happy to agree. It is exactly because the world is complex that commissioners shouldn’t meddle around specifying activities. Rather, outcomes-based approaches incentivise providers to innovate. This ensures that those with the best information (the providers and those they work with) are freed up to achieve the very best results. So SIBs have complexity thinking at their heart.
But this complexity is wrapped in conformity because SIBs also demand that the outcomes by which success is judged must be measurable, attributable and priceable. It is this aspect which renders SIBs ultimately a mechanistic, deterministic tool rather than one that is truly based on the insights of complexity.
So SIBs simultaneously both embrace and reject a complex view of the world and it is this inherent inconsistency, I believe, that makes them fatally flawed. No matter how you design them, at some point, the top-down, linear logic of outcomes-based payments has to meet the bottom-up, emergent reality of the world. And reality always wins.
One response to this challenge is to invest increasing amounts of time, energy and intelligence attempting to design ever-better metrics and mechanisms. I’m afraid these costly efforts are doomed to fail, as the complexity remains regardless. Another response is to point out that at least SIBs are better than the bad old days when contracts specified activities. While this may be true, it is a particularly unsatisfying answer. Can’t we find a better way?
The most promising ideas shared at the conference “adapted” the SIB model so that it was based less on defined payments for defined outcomes and more on providing a framework within which collaboration and innovation can be fostered. In other words, scaling back the top-down deterministic element sometimes to the point of non-existence. As Scott Kleiman from Harvard’s Government Performance Lab (formerly known as the SIB Lab) admitted — when people come asking for help implementing SIBs he usually recommends something else.
While SIBs were never “bonds” in any meaningful sense of the word, once the outcomes payments and equity components are removed I think we can safely drop the quasi-financial nomenclature entirely. Perhaps a new label such as social impact partnerships (SIPs) would make more sense — or put more simply — collaborative working.
Call me old-fashioned, but the best way of financing these would be through grants, where the performance conversation is just that, a conversation about performance, rather than a spreadsheet, and it is assumed that those involved are intrinsically motivated to make the world a better place. There is already a good deal of thinking on models like this and examples of how they can work in practice. See, for example, Dr Toby Lowe’s work on commissioning in complexity.
My suggestion is that at the Social Outcomes Conference 2020 we dedicate two sessions to SIBs as an interesting, but largely historical detour, and spend the remaining 21 exploring how we can achieve better outcomes in a complex world.
