What Are Social Impact Bonds?

An Innovative New Financing Tool for Social Programs

Jitinder Kohli, Douglas J. Besharov, and Kristina Costa answer basic questions about Social Impact Bonds—what they are, where they can be most useful, and how they differ from traditional government contracts.

Part of a Series
Inmates at the Department of Youth Services juvenile boot camp wait to go outside for physical training in Prattville, Alabama. Social Impact Bonds have enormous potential in areas of social policy such as homelessness, juvenile delinquency, preventive health care, and workforce development. (AP/Rob Carr)
Inmates at the Department of Youth Services juvenile boot camp wait to go outside for physical training in Prattville, Alabama. Social Impact Bonds have enormous potential in areas of social policy such as homelessness, juvenile delinquency, preventive health care, and workforce development. (AP/Rob Carr)

Download this issue brief (pdf)

Read the full brief in your web browser (Scribd)

See also: Social Impact Bonds 101 by Jitinder Kohli, Douglas J. Besharov, and Kristina Costa

This is the first in a series of issue briefs that looks at Social Impact Bonds and their value to government agencies. Subsequent pieces will focus on getting the SIB agreement right; models for SIBs and their long-term potential; defining and measuring outcomes for SIBs; and appropriate roles for government agencies in the SIB process.

Below we answer basic questions about Social Impact Bonds—what they are, where they can be most useful, and how they differ from traditional government contracts. We also look at some of the issues government needs to consider before beginning Social Impact Bonds. These arrangements show real potential to help solve difficult social problems and give taxpayers better value for their money.

What are Social Impact Bonds?

Social Impact Bonds are a new and innovative financing vehicle for social programs that flip traditional government funding structures on their head. Instead of paying upfront for a proscribed set of services, SIBs allow government to focus funds on approaches that work—without paying a dime if agreed-upon outcomes are not achieved. SIBs work by bringing together government agencies, social service providers, and philanthropically minded financiers to achieve better results for people receiving social services and for the taxpayers funding those services.

The Social Impact Bond concept is relatively straightforward: Government agencies define an outcome they want to accomplish and agree to pay an external organization a sum of money if the external organization achieves that outcome. This unusual mechanism promotes innovation in public services by putting taxpayer dollars toward the most effective approaches. This is markedly different from normal funding arrangements for social programs, in which agencies typically commit to funding activities regardless of the outcome.

While Social Impact Bonds are still in their infancy, the concept has enormous potential in areas of social policy such as homelessness, juvenile delinquency, preventive health care, and workforce development. In particular, SIBs could become a key vehicle for funding prevention initiatives designed to save government money over the long term. Early interventions to reduce homelessness or smoking, for example, could lead to considerable reductions in Medicare and Medicaid spending—not through program cuts but through lower hospitalization rates.

A definition of Social Impact Bonds

Before we delve deeper into the concept, let’s start with a simple definition of Social Impact Bonds:

An arrangement between one or more government agencies and an external organization where the government specifies an outcome (or outcomes) and promises to pay the external organization a pre-agreed sum (or sums) if it is able to accomplish the outcome(s).

In addition, SIBs require:

  • Government to place few, if any, controls on the way that the external organization accomplishes the outcome
  • Government to cooperate with the external organization so that it is able to take the actions necessary to achieve the outcome—for example, by ensuring access to relevant data
  • A clearly defined population and clarity on what a “successful outcome” would be

Some things to note:

  • All payments are contingent on the outcome being achieved. If outcomes are not achieved, the government pays nothing. Hence, risk is transferred from the government to the external organization or its investors.
  • The crux of the relationship is between government and the external organization committed to accomplishing the outcome. While there may be other players present (as discussed below), they are not essential to the concept, and they do not have a direct relationship with government.
  • While Social Impact Bonds are likely to be particularly useful in areas where accomplishing outcomes results in direct savings for government, not all Social Impact Bonds have to result in government savings.

Where are Social Impact Bonds most likely to be useful?

The concept is still new and there is a great deal to learn. But the most likely early applications will be in areas where:

  • Government agencies believe they will save money as a result of the outcome being accomplished and those savings are likely to accrue over a reasonably short timeframe.
  • The outcomes are observable and measurable within perhaps three to eight years of the agreement, and government payments flow in that timeframe so that investors are not asked to tie up their funds for too long.
  • It is possible to observe and measure whether outcomes are achieved in an objective rather than subjective manner.
  • Research shows there are effective interventions to achieve the outcome and these interventions cost less than the government is willing to pay for accomplishing the outcome. This is necessary so the external organization has a degree of confidence that it can achieve the outcome in a cost-effective manner and therefore receive enough money to repay investors.
  • There are few negative consequences to the beneficiaries of the service if the external organization is unable to achieve the outcome and seeks to discontinue its services. In the Peterborough example, the services provided under the Social Impact Bond are new and not typically offered to inmates in other prisons. If they were to be discontinued, therefore, the inmates would be no worse off than those from other prisons.
  • The external organization is able to influence the outcome significantly (i.e., the effectiveness of its efforts is not likely to be heavily affected by circumstances outside its control).

Here are some specific areas that governments in the United States are beginning to explore for Social Impact Bonds:

  • Reducing recidivism, as in the Peterborough example
  • Reducing homelessness
  • Preventive public health (e.g., reducing smoking rates, obesity, etc.)
  • Helping people re-enter the workforce and workforce development

There is particular interest in using Social Impact Bonds in areas where successful preventive services could reduce future public expenditures. The Peterborough SIB on recidivism is built on this premise.

We will be publishing a subsequent brief that focuses specifically on program areas where SIBs might be most useful.

What are the advantages of Social Impact Bonds?

Governments around the world are getting interested in Social Impact Bonds. So what’s in it for them? We believe there are four key advantages:

  • Genuinely directing resources to outcomes: While government agencies are increasingly aware of the need to focus resources on outcomes, the reality is that most government programs still fund activities upfront. If these activities fail, taxpayer dollars are still spent. But Social Impact Bonds are different: Government only directs resources toward things that work.
  • Overcoming government silos, which should improve effectiveness: Government agencies find it hard to think beyond the silos of different programs. But the best way to reduce recidivism, for example, might be some combination of support to help offenders find jobs and housing, alongside overcoming their drug dependency or anger issues and developing a deeper understanding of crime’s consequences on victims and society. The optimal mix of support may vary significantly for each individual offender. Government programs usually don’t work that way, however. Instead, separate programs will likely focus on each of the different components of a successful strategy to reduce re-offending. As a result, those who need support to reduce re- offending may receive patchy help from one or more of these programs, each of which has little incentive to work with the others. But with a Social Impact Bond focused on reducing recidivism the external organization has a strong incentive to coordinate these different approaches in order to successfully achieve the outcome.
  • Allows funding to shift toward effective approaches: Governments find it hard to move money from ineffective programs to those that work well. Sometimes this is because government agencies are poor at measuring whether programs work, focusing their energy on disbursing funds instead of measuring impact. But even where that data exists, the political barriers to realigning resources can be tremendous. Even programs that are ineffective have vocal supporters and these supporters are often good lobbyists, focusing their influence on senior officials in the executive branch or on appropriators in the legislative branch who determine budget priorities. But with a Social Impact Bond, the external organization has strong incentives and sufficient freedom to direct funds to approaches that work—and the process of doing so is depoliticized.
  • Scaling innovations: While government agencies often talk of the importance of scaling successful innovations, in reality innovative approaches that work at the small scale often find it hard to attract the government funding needed to grow. Sometimes this is because government agencies don’t always know what works, and sometimes it’s because funding streams are overly prescriptive and only provide support to approaches that meet tightly defined criteria. But with Social Impact Bonds the external organization is encouraged to identify approaches that have worked elsewhere and look to scale them up or expand them into new regions.

Challenges for government

Social Impact Bonds are a new concept, and there is considerable interest from federal, state, and local governments, particularly because the SIBs seem particularly suited to tight fiscal times.

But governments need to consider a few issues:

  • Defining outcomes is not easy: Defining a meaningful outcome and determining the correct “price” to pay if it is accomplished is hard, and will require a set of skills that government agencies may not readily possess. It’s also important for outcomes to be objectively and rigorously measurable to avoid disagreement between government and the external organization about whether the outcome has been achieved and payment is due. Often it will be necessary to appoint a third-party organization that determines whether the outcome has been achieved—and specifies the methodology it uses to determine this.
  • Avoid worrying about investor return: Government agencies should set a payment schedule for the outcome based on what they think the outcome is worth to them and society as a whole. But government will have a natural tendency to tie payment to what it considers to be acceptable returns to the external organization or its investors. This is not the right way forward. For the model to work well, governments should focus on what they are willing to pay for the outcome and resist the tendency to judge what they deem to be an acceptable return for investors.
  • Clarity of roles: Social Impact Bonds require government agencies to take the lead in negotiations with external organizations on the correct outcome and price. But government agencies must leave day-to-day decisions to the external organization or its partners. In particular, government agencies need to be willing to cede considerable power over how the external organization achieves the outcome. If government tells the organization how to do the job it defeats the flexibility and innovation inherent in SIBs.
  • Building trust is essential: These are complex arrangements and require considerable trust between government agencies and external organizations. Governments will need to ensure that procurement processes are not so formal that they constrain the ability to build trusting relationships with external organizations, while at the same time delivering value for taxpayer dollars. Government agencies, for example, will want to meet with key people from external organizations as part of the decision-making process.
  • Funding promises must be firm: Social Impact Bonds are multiyear arrangements where government promises payment in future years if the external organization achieves the outcome. But appropriations processes can make it difficult for government to make guaranteed promises about spending in a future year. Indeed, a change in administration or composition of the legislature can take place between the commitments government makes and its payments. Government agencies will need to find ways to offer external organizations firm guarantees that payments will be made if outcomes are accomplished.
  • Define exit arrangements: The external organization has strong incentives to do its best to achieve the outcome, but the payment’s “all or nothing” nature means the external organization also has an incentive to walk away if its efforts are falling short and it thinks the chances of accomplishing the outcome are very low. Government will therefore need to negotiate mechanisms for an orderly termination of the arrangement if necessary. It will also need to make sure that in the event of termination those who received services are no worse off than they would have been. If government agencies do not realize termination is a real possibility there is a risk they will be tempted “bail out” external organizations by renegotiating agreements.
  • Benefits can accrue to different agencies and at different levels of government: Take homelessness, for example. If there is a reduction in chronic homelessness in a community, the local housing department could see considerable savings, but so could the federal government in programs such as Medicaid or the Veterans Health Administration. Payments to external organizations, therefore, should ideally be funded by a combination of agencies. That gives rise, however, to significant logistical challenges in government that may need to be overcome.

It is important for government to recognize that Social Impact Bonds are not a panacea. They could be transformative for many social programs, but they also will be inappropriate in many areas.

Isn’t this just a performance contract?

Social Impact Bonds differ from traditional government performance contracts in two key ways. First, most performance contracts contain a significant fixed payment from government to the contractor for the activities alongside a bonus for achieving results. But in a SIB the upfront payment is zero, and any actual payment hinges on accomplishing the outcome.

A second key difference is that in most performance contracts the government places major constraints on how contractors operate. For instance, a government contract recipient may be required to use only approved materials and methods. The Social Impact Bond model doesn’t contain these prescriptive requirements, and the external organization has considerable freedom on how to achieve the outcome.

Does a SIB have to be done the way it’s being done in Peterborough?

No. There are a number of variations to the Peterborough SIB that remain consistent with the Social Impact Bond model:

  • No need for external investors to be present: If the external organization is able to raise the funds necessary for the intervention from its own balance sheet rather than from investors, that is consistent with the SIB approach.
  • No need for a separate set of providers: The external organization can choose to be one of the service providers, or even the sole provider.
  • Targeting more than one outcome: A number of outcomes could form the basis of the agreement and the government could promise differing payments depending on which are accomplished. For instance, the agreement could target reductions in re-offending and increases in employability for the same population and have different payments associated with both.
  • External organizations could be for-profit: The external organization in Peterborough is a nonprofit, as are those providing services. If these organizations had a profit motive that would still constitute a Social Impact Bond.

Variations to the Social Impact Bond exist that do not result in a SIB per se but that improve on traditional contract arrangements:

  • Guaranteed payment: The government could make a relatively modest level of guaranteed payment to the external organization even if outcomes are not achieved.
  • Obligation to continue services: The external organization could be required to continue to provide services through the term of the arrangement or for a set period of time even when it believes it has a low chance of achieving the necessary outcomes.
  • Greater government control: The government could place more controls on how the external organization achieves the outcome (e.g. by insisting that it only deploy techniques that government has vetted).

The agreement is not a Social Impact Bond if any of these three variations are made, and it becomes more like a performance contract. In some cases, that will be a more appropriate arrangement, as in cases where government feels it needs more control over how outcomes are accomplished.


Our country has real opportunities to better solve social problems through partnerships between the government, nonprofits, and the private sector. Social Impact Bonds are among the newest and most innovative ways to create those partnerships. Although SIBs are in their infancy, there is tremendous excitement in some state and local governments, in the federal government, and among foundations and nonprofits about the potential impact of this new model for funding social programs.

Those interested in solving egregious social issues are right to be excited about these unusual public-private funding mechanisms. Social Impact Bonds could significantly improve the quality of public services, save taxpayer money, and offer new approaches to providing social services without requiring government to assume all of the financial risk. But most importantly, SIBs could help give taxpayers significantly better returns for their investments.

The next issue brief in this series will explain what terms are required for a Social Impact Bond contract to work correctly—and what kinds of provisions should be kept out of SIB agreements.

Jitinder Kohli is a Senior Fellow with the Doing What Works project at the Center for American Progress.

Douglas J. Besharov is a professor at the University of Maryland School of Public Policy and a senior fellow at the Atlantic Council.

Kristina Costa is a Research Assistant at the Center for American Progress.

The Center for American Progress’s work on Social Impact Bonds is supported by the Rockefeller Foundation.

The authors are extremely grateful to those who reviewed earlier drafts of this piece and provided insightful comments. In particular, they would like to thank Kristin Giantris, Gary Glickman, Steve Goldberg , Sarah Henderson, Kippy Joseph, Ben Jupp, Ben Kershaw, Justina Lai, Kristin Misner, Tracy Palandjian, William Pinakiewicz, Jim Robinson, Sonal Shah, Jennifer Talansky and Marta Urquilla.

Download this issue brief (pdf)

Read the full brief in your web browser (Scribd)

See also:

The positions of American Progress, and our policy experts, are independent, and the findings and conclusions presented are those of American Progress alone. A full list of supporters is available here. American Progress would like to acknowledge the many generous supporters who make our work possible.


Jitinder Kohli

Senior Fellow

Kristina Costa

Senior Fellow

Explore The Series