“For a not-for-profit it’s the equivalent of venture capital,” said Sir Ronald Cohen, chairman of the Global Social Impact Investing Steering Group, about impact bonds in his keynote address at a recent event at the Brookings Institution. Impact bonds combine results-based financing and impact investing, where investors provide upfront capital for a social service and government agencies, or donors, agree to pay investors back based on the outcomes of the service. At their best, they could allow for innovation, encourage performance management and adaptability, promote learning through evaluation, and create a clear case for investing in what works. However, impact bonds thus far have had immense transaction costs and there are risks that poor execution of the impact bond mechanism could have negative consequences for beneficiaries.

It has been six years since the first impact bond was implemented in March of 2010, and the field is beginning to move from an exploratory stage to looking at systemic change, as Tracy Palandjian, CEO and co-founder of Social Finance U.S. described. The event, “The Global Potential and Limitation of Impact Bonds,” served as a point of reflection for stakeholders at this pivotal stage of the field, bringing together over 500 individuals in the room and on the webcast, including practitioners developing impact bonds around the world. While context matters, there were notable similarities in the motivations and challenges across regions.

In our presentations of our research and subsequent panels, we focused on the potential value and challenges of combining results-based financing and impact investing through an impact bond. Shri Naveen Jain, mission director of the National Health Mission of Rajasthan, India, who is working to develop an impact bond for maternal and child health services across his entire state, pointed out that the value of a results-based financing contract to him was in the added transparency it provides—the government is able to see what they are paying for, keep service providers accountable, and incentivize providers to achieve better outcomes. Louise Savell, a director at Social Finance U.K., the entity that first put impact bonds on the map, explained that results-based financing contracts are often arranged such that only one portion of the contract is based on results. This, she explained prescribes a model and does not allow for flexibility; furthermore, it forces service providers to bear a significant risk. Impact bonds allow for the entirety of payments to be based on results, which gives the provider full flexibility (at least in theory), but puts the risk of service performance on the investor. The shift of risk to investors could be particularly useful for service delivery in conflict affected areas, where donors are often highly concerned about how money will be used, mentioned Francois de Borchgrave, co-founder and managing director of Kois Invest, who is working on an impact bond with the International Rescue Committee of the Red Cross. The panelists also emphasized that impact bonds are more powerful than results-based financing contracts alone because, if successful, they pay real financial returns to investors. This draws a great deal of attention from policymakers and the public, and the added scrutiny helps in making the investment case for preventive interventions highly explicit. Mayor Ben McAdams of Salt Lake County, Utah said that “data and evidence is bridging a partisan divide” in his state—when the case for investment is clear, policymakers from both sides of the aisle are willing to invest. Impact bonds do not necessarily add value by increasing the total amount of funding available for social services, because investors are repaid if outcomes are achieved. Rather, impact bonds could help increase the outcomes achieved with given funding.

Overall there was agreement that impact bonds have enormous potential to lead to more outcome-focused financing that focuses on preventive interventions and incentivizes collaboration. However two critical considerations for the use of impact bonds arose throughout the day.

The first consideration discussed was whether or not impact bonds can support innovation or scale. As found in our first report, impact bonds have been relatively small in scale in terms of capital and beneficiaries. The average upfront investment in impact bonds to date is $3.7 million, reaching an average of 1,900 beneficiaries. They also have not, on average, focused on particularly innovative interventions—in fact they have almost all had a relatively strong base of evidence behind them. Views on the panel differed on whether the uses of impact bonds could be expanded—if they could be used for highly innovative pilot programs or proven large scale programs. One perspective was that impact bonds could indeed provide seed capital to test new ideas for service delivery. This would require investors who are willing to test not only the innovation but also this relatively new financing mechanism. Given the high transaction costs that impact bonds entail, however, this may not be the most efficient use of resources. Impact bonds could also reach more beneficiaries per transaction (greater scale) with changes in public procurement and the creation of markets for tradeable impact bond assets. Government can play a role in facilitating larger impact bonds by creating central government outcome payment funds, providing tax breaks for investment in impact bonds, and enabling the development of investment vehicles, all of which are being implemented in the U.K. Impact bonds could also help effective social services reach scale by encouraging government to fund programs at scale after the impact bond is over or by improving data use and performance management in government-funded services broadly.

A second, and related, discussion happened around evaluation methodology—which may differ depending on whether the impact bond is intended to test an innovative intervention or scale an intervention already backed by significant evidence. The “gold standard” randomized controlled trial (RCT) is the only methodology that eliminates the possibility that impact could be attributed to something other than the intervention, though the majority of impact bonds thus far use evaluation methodologies that are less rigorous. The panelists explained that it is important, however, to consider the status quo—currently, less than 1 percent of U.S. federal spending on social services has been shown to be effective. The same is true in low- and middle-income countries, where there are relatively few impact evaluations given the number of interventions. At the end of the day, the government agency acting as the outcome funder must decide on the importance of attribution to trigger payment through the impact bond in view of the already available evidence of program effectiveness and weigh the criticism that might ensue in the absence of a valid counterfactual.

Though impact bonds are a potentially useful tool in the toolbox of many financing mechanisms, there are some significant constraints to their implementation. The biggest barrier to impact bonds and other results-based contracts is the administrative hurdle of contracting for outcomes. Peter Vanderwal, innovative financing lead at the Palladium Group, and Caroline Whistler, co-president and co-founder of Third Sector Capital Partners, both stated that governments often are unable or do not know how to contract for outcomes, and there is a need to invest in their capacity to do so. Appropriation schedules are part of this challenge, governments are often not allowed to appropriate for future years. When an audience member asked how we go about changing the culture in government to one of contracting for outcomes, Mayor McAdams answered that impact bonds may have a contagious effect—contracting for outcomes will be the expectation in the future. Additionally, the transaction costs of establishing the partnership are large relative to other mechanisms, though they may be worthwhile. Jim Sorenson, of the Sorenson Impact Center, pointed out that service provider capacity and data collection systems could be barriers to the development of future impact bonds. There is also still a long way to go in developing outcome measures and in particular in calibrating those outcome measures to low- and middle-income countries.

The influence that impact bonds have on the provision of quality services globally depends on the quality of implementation. With a rapidly growing market, there will inevitably be “bad” impact bonds in the future. To ensure that impact bonds are used as effectively as possible, governments and the research community have a pivotal role to play in asking the right questions: Will a results-based contract help improve outcomes in this particular case? What should the outcomes be to avoid perverse incentives or potentially negative externalities? And would an impact bond structure add value? 

“This is about governments using data for performance rather than compliance” was a resounding message coming out of the U.S. Department of Education’s conference on June 10 on the use of Pay for Success contracts in education. These contracts, known globally as social impact bonds, continue to be at the forefront of global conversations about results-based financing mechanisms, and have garnered significant momentum this week with passage of the Social Impact Partnerships for Pay for Results Act in the U.S. While limitations certainly exist, their potential to revolutionize the way we fund social projects is tremendous.

A social impact bond (SIB) is a set of contracts where a government agency agrees to pay for service outputs or outcomes, rather than funding defined service inputs, and an investor provides upfront risk capital to the service provider. The investor is potentially repaid principal and interest contingent on the achievement of the predetermined outputs or outcomes.

In our research on impact bonds at the Center for Universal Education, we have analyzed the use of SIBs for education in the U.S., other high-income countries, and low- and middle-income countries. Practitioners in each of these contexts are having far more similar conversations than they may realize—all are united in their emphasis on using SIBs to build data systems for performance. There is tremendous potential for lessons learned across these experiences and across the broader discussions of results-based financing mechanisms for education globally.

Current SIBs for education globally

There are currently five SIBs for education worldwide: two in the U.S. for preschool education, one in Portugal for computer science classes in primary school, and one each in Canada and Israel for higher education. In addition, a number of countries have used the SIB model to finance interventions to promote both education and employment outcomes for teens—there are 21 such SIBs in the U.K., three in the Netherlands, and one in Germany. There is also a Development Impact Bond (DIB), where a donor rather than government agency serves as the outcome funder, for girls’ education in India. The Center for Universal Education will host a webinar to present the enrollment and learning outcomes of the first year of the DIB on July 5 (register to join here).

U.S. activities to facilitate the use of SIBs for education

At the June 10 conference at the Department of Education, the secretary of education and the deputy assistant to the president for education said that they saw the greatest potential contribution of SIBs in helping to scale what works to promote education outcomes and in broadening the array of partners involved in improving the education system. Others pointed out the value of the mechanism to coordinate services based on the needs of each student, rather than a multitude of separately funded services engaging the student individually. In addition to using data to coordinate services for an individual, participants emphasized that SIBs can facilitate a shift away from using data to measure compliance, to using data to provide performance feedback loops.

The interest in data for performance rather than compliance is part of a larger shift across the U.S. education sector, represented by the replacement of the strict compliance standards in the No Child Left Behind Act of 2002 with the new federal education funding law, the Every Student Succeeds Act, signed into law in December of 2015. The law allows for federal outcome funding for SIBs in education for the first time, specifically for student support and academic enrichment programs. The recently passed Social Impact Partnerships for Pay for Results Act also allows for outcome funding for education outcomes. The Department of Education conference explored potential applications of SIBs across the education sector, including for early home visiting programs, programs to encourage completion of higher education programs, and career and technical education. The conference also analyzed the potential to use SIBs for programs that support specific disadvantaged populations, such as dual language learners in early education, children of incarcerated individuals, children involved in both the child protection and criminal justice systems, and Native American youth. Overall, there was a focus on areas where the U.S. is spending a great deal on remediation (such as early emergency room visits) and on particular levers to overcome persistent obstacles to student success (such as parent engagement).

To help move the sector forward, the Department of Education announced three new competitions for feasibility study funding for early learning broadly, dual language learners in early education, and technical education. The department is also facilitating connections between existing evaluation and data system development efforts and teams designing SIBs. The focus on early childhood development by the Department of Education is reflective of the national field as a whole: Programming in the early years is becoming a particularly fast-growing sector for SIBs in the U.S. with over 40 SIBs feasibility and design stages.

SIBs for education in low- and middle-income countries

There is only one DIB for education in low- and middle-income countries; however, there are a number of SIBs and DIBs for education in design and prelaunch phases. In particular, the Western Cape Province of South Africa has committed outcome funding for three SIBs across a range of health and development outcomes for children ages 0 to 5.

Though the number of impact bonds may be relatively small, a significant amount of work has been done in the last 15 years in results-based financing for education. The U.K. Department for International Development (DfID), the Dutch Ministry of Foreign Affairs, the Asian Development Bank, the World Bank, the Global Partnership for Output-Based Aid, and Cordaid had together funded 24 results-based financing initiatives for education as of 2015. Of particular interest, DfID is funding results-based financing projects through a Girls Education Challenge and the World Bank launched a new trust fund for results-based financing in education in 2015. As with impact bonds in the U.S., a primary aim of results-based financing for education in low- and middle-income countries is to strengthen data and performance systems. Early childhood development programs and technical and vocational and training programs have also been identified as sub-sectors of high potential. Here are a few final takeaways for those working on results-based financing for education in low- and middle-income countries from the U.S. Department of Education conference:

1. The differences between the No Child Left Behind Act and the Every Student Succeeds Act should be analyzed carefully to ensure other data-driven education performance management systems promote both accountability and flexibility.
2. In building data systems through results-based financing, ensure services can be coordinated around the individual, feedback loops are available for providers, and data on early education, child welfare, parent engagement, and criminal justice involvement are also incorporated.
3. There are potential lessons to be learned from the U.S. Department of Education’s effort to conduct more low-cost randomized control trials in education and the U.S. Census Bureau’s data integration efforts.
4. SIBs provide an opportunity to work across agencies or levels of government in education, which could be particularly fruitful in both low- and middle-income countries and the U.S.

As the global appetite for results-based financing continues to grow and new social and development impact bonds are implemented throughout the world, we’ll have an opportunity to learn the true potential of such financing models.