How can we use tech to verify outcomes for impact bonds?
Frontier tech offers an opportunity to measure whether a project has achieved impact, but there are barriers to adoption.
By Blair Seiler, Associate at Social Finance
Since October 2020, we have been developing a pilot, alongside the Frontier Tech Hub and the Foreign, Commonwealth and Development Office (FCDO), to explore using frontier tech to evaluate whether a project has achieved impact, identifying where it can increase the speed, trustworthiness, and cost-efficiency of verifying social outcomes. We chose to zoom in on the use case of outcomes verification for impact bonds (read an introduction to the project).
In this blog we discuss key opportunities to use tech for impact verification, the barriers to adoption, and how transformational change can be achieved. In the long-run, we hope our learnings will progress the conversation around how to drive effectiveness and adaptability for international development programmes.
How can technology be used to verify outcomes in impact bonds?
Development impact bonds (DIBs) are outcomes-based contracts where a payer — usually a government or donor — commissions services, but pays only for actual outcomes achieved. A social investor provides the upfront capital, taking on the risk of low success in exchange for a possible return on their loan. One scaling issue for DIBs is that the steps from results to payment can be time-consuming and expensive. Verifying hundreds — or in some cases, thousands — of individual outcomes is a largely manual process, performed through routes like phone calls, surveys, and cross-referencing records.
Social Finance set out to understand what types of technologies could help to transform the results to payment cycle. Here’s what we learned:
- There is huge variety in the technology being used to verify social outcomes… such as machine learning, sensors, healthcare and satellite imaging tech, mobile apps, blockchain-based platforms and predictive analytics. Many solutions combine multiple technologies and tend to be software-focused with some hardware elements.
- Few of these technologies have been developed for scale. The majority need to be tailored, often extensively, for each use case. Much of the innovation in the sector is being led by development-focused nonprofits and service providers who have adopted agile and low-cost solutions to fit small budgets and short timeframes.
- Certain outcomes are better suited to verification by technology. These tend to be those that are simple and observable, or require small human inputs. There is a concentration of verification solutions for infrastructure (e.g. sanitation facilities) and health outcomes.
How can impact bonds leverage these technologies?
To understand how impact bonds might leverage these technologies to verify outcomes, we spoke to a number of DIBs in development and to others that have recently launched. There were two promising paths forward that we uncovered:
Fund technical capacity of current service providers: A number of DIBs under development are working with service providers that rely heavily on innovative technologies. For example, three of the DIBs will fund healthcare providers who use mobile-based health screening and data collection technologies to measure health outcomes — these rely on a combination of imaging technologies, social media data, and mobile and tablet applications. Another three DIBs are exploring sensors in community infrastructure like water, sanitation, hospitals and home devices, to measure impact on communities.
Tech-savvy service providers have the advantage of expertise on their beneficiaries and their “problem,” and have locally adapted solutions. Where possible, their existing solutions may be ideal platforms to build out more efficient and comprehensive outcomes verification functionalities. Service provider-led solutions are also likely to be more sustainable, although they are less easily scaled across an issue area.
Create partnerships with tech innovators: Partnerships with research labs and tech startups looking to build on first use cases have been an interesting area where new solutions have arisen. These have the opportunity to bring innovation to impact bonds, at a cost that may be subsidised by donor funding, academia or venture funds.
What stands in the way of adoption?
Given the amount of innovation in the sector, we want to highlight the main obstacles to adoption we believe need to be addressed to reach higher adoption. These are below.
1. Clarity of stakeholder incentives
There are advantages for each DIB stakeholder in adopting verification technology: trust and accountability for funders, speed of impact measurement and reduced costs for investors, and the ability for real time learning and adaptation for service providers. But these benefits are not evidenced and documented broadly enough to measure against the perceived risks and costs.
For example, DIB service providers mentioned transition costs and the risks of increased transparency as barriers, while DIB funders believed using emerging tech could be unpredictable. This was described by an interviewee as “risk laid upon risk, a technological innovation laid upon a financial innovation.” There are a few trailblazers, but more evidence of the incentives are needed before all stakeholders can buy in.
2. High resources to adoption (cost/time)
Our interviewees reported that the wide availability of open-source models for verification, and relatively cheap hardware like sensors should make it affordable and quick to build these verification solutions. So why are impact bonds finding it so expensive?
In short, there is limited funding to scale suitable solutions, and processes extend timeframes.
For small tech providers, donor funds might be the most likely payer — but these require complex reporting and procurement structures, and often the contract sizes are many orders of magnitude too small to scale a solution. Larger tech providers may offer ‘off-the-shelf’ solutions. But these are often less suitable: developed in global tech hubs, and not adapted for low-resource settings where wifi connectivity and the availability of mobile phones is not guaranteed. They may require costly tailoring to local use cases, and long-term maintenance that service providers can’t sustain themselves.
Processes that are key in social outcomes programmes like procurements, legal contracting, and staff training can take years, a timeline at odds with the shorter timeframes (3–5 years) of a typical DIB. In response, we’ve seen service providers opting to build lean tech solutions in-house and developed iteratively for their project needs, rather than commissioning out to a tech company.
3. Data ethics and privacy
This takes time to understand how to get right. Development projects are seeking to understand: who is the right party to hold data, and when can it be a tech company? In some contexts, country governments will also have preferences or regulations on where to hold the data in a given programme.
Many data privacy issues stand to be solved by emerging technologies, particularly ones that remove manual elements of data entry such as imaging technologies, or store data with high fidelity, like blockchains. Yet there is a perception that some of these solutions are still not fully adopted in western markets — this breeds the fear that first use with developing country populations could be harmful and result in unexpected exposures of sensitive data.
In reality, many of these technologies have been tested extensively, notwithstanding their “emerging” status and low number of use cases in the social and charitable services markets. Therefore, the problem to solve may be around addressing cultural and organisational resistance to changing the way data is collected, due to perceived risk.
How do we overcome these barriers, to set the scene for transformational change?
There is huge potential for emerging tech to bring efficiency, accuracy and trustworthiness to assessing impact — particularly for the impact bond market. The tech solutions themselves are diverse, innovative and rapidly developing. The biggest roadblock to transformational change is not in the technology but in adoption, requiring better clarity on incentives, lower resources to adoption and more considered data ethics processes.
We are looking to design a series of experiments to explore these barriers. We believe the most important aspect will be demonstrating incentives for the key DIB partners: funders, investors, service providers and even tech providers. Evidencing the business case, costs and incentives, the challenges, and the solutions to overcome them will help future projects navigate this new territory. We will need to begin by leveraging what is available: the huge amount of progress in the sector on the technical side, excitement from funders and investors, and the deep experience of service providers regarding implementation and the needs of their users.