How this Works > Prototypes

Social Impact Guarantees Could Enable Pay for Success Contracting to Scale More Rapidly

By George Overholser

Recently, a senior government official leaned over the table, looked me in the eye, and asked, “Is this going to get any easier?” “Yes, Pay for Success will get easier,” I responded. “Much easier, I’m willing to bet. But only if we continue to innovate.”

One innovation we are particularly excited about is something we are calling the social impact guarantee. If a government-backed social program fails to achieve social impact, the government gets its money back. And for service providers that don’t want to take on the risk of providing their own money-back guarantee, private funders can offer social impact guarantee financing. That way, if the social service provider is called upon to pay back the government, social impact guarantee funders will step in to write the check. In contrast with the social impact bond (described in Tracy Palandjian’s chapter in this volume), where private funders write checks at the beginning and the government (potentially) writes checks at the end, the social impact guarantee has the government writing checks at the beginning and private funders (potentially) writing checks at the end. You might say that a social impact guarantee is a social impact bond in reverse.

Both of these approaches reach a similar place. The government pays only if social impact is achieved. And both use private financing to offload performance risk from vulnerable service providers. But in many ways, the social impact guarantee approach can be simpler and philosophically more intuitive than the social impact bond.

To understand why, consider that one of the reasons that social impact bonds can be so difficult to implement is that the government is not accustomed to contracting for social services in a multi-year or contingent way. Even if government officials would like to take on multi-year Pay for Success obligations, laws and regulations may be a barrier. There are workarounds — waiver processes, legislative overrides, creation of special new spending streams with special use rules, elaborate sinking funds, and the like. But the workarounds are technical, expensive, uncertain, and time-consuming. Indeed, this has been a significant impediment to the growth of Pay for Success contracting.

A second technical problem with Pay for Success/social impact bonds is called “double capitalization.” Originally, the idea was to reduce stress on government cash flows by providing private upfront social impact bond financing. But in practice, governments have needed to place capital into escrow accounts right from the start, so that social impact bond lenders feel confident that the government has enough cash lined up to make good on its Pay for Success payment obligations. This double capitalization has been necessary to give funders confidence, but it is an inefficient way to scale Pay for Success contracting more broadly.


With the social impact guarantee, the model is flipped: Instead of the government, it is private funders who make a contingent promise to pay in the future. This allows the government to purchase social services in a way that more closely resembles its usual pay-as-the-work-is-done timing. Then, if social impact targets are missed, private funders write a check back to the government. For example, a group of private social impact guarantee funders might obligate themselves today to write a check three years from now, but only if a program fails to reduce the incarceration rate. The good news is that multi-year contingent promises are something private funders do all the time. No waivers needed, no new government spending streams with special rules. It’s a lot simpler.

Technically, the government would implement social impact guarantees by inserting a penalty provision into its social services contract that makes clear to the organization awarded government funding to provide social services that if those services do not work, the government will have to be repaid some or all of its funds. Then the contract would specify impact metrics that determine what constitutes success and what would trigger the penalties. Luckily, penalty provisions are already common practice, and specifying impact metrics is no different from what is already done under the Pay for Success/social impact bond model.

The other good news is that the social impact guarantee does not require the government to place capital redundantly into sinking funds and the like. Instead, the government’s money is used productively from the start. The problem with traditional penalty provisions is that they can disrupt or even bankrupt services providers who have to come up with money to pay back the government after they have already spent their funding on service delivery. That’s where, in the social impact guarantee approach, private investors step in. They offer something akin to an insurance policy or a letter of credit, which means they (not the service provider) will ultimately be on the hook to cover the penalty. The insurance policy or letter of credit is a contingent promise to write a check in the future rather than provide an immediate outlay. In other words, there are no escrow accounts with money that simply sits there. Bottom line: The social impact guarantee would have less “double capitalization” than the social impact bond, and that makes it inherently less expensive.

Philosophically, I have felt that the social impact bond places too much emphasis on borrowing and not enough on insurance. Indeed, the central role of a social impact bond (and certainly of a social impact guarantee) is insurance. It insures against the risk of the government’s allocating precious taxpayer money to programs that do not work. And it insures vulnerable nonprofit service providers from the financial peril of potentially never being paid.

Currently, social impact bonds are positioned as a lending product, prompting people immediately to ask whether the government really needs to borrow more money, especially when the investment returns typically demanded by investors for investing in social impact bonds are so much higher than, say, municipal bonds. With the social impact guarantee, everything begins to make more sense. This is not an interest payment; it is an insurance premium. And it is easy to explain how paying a five percent premium for something that might otherwise cost 100 percent is a good deal — especially in a world where so many social programs, when rigorously evaluated, are revealed to have been ineffective.


Of course, as with any innovation, the social impact guarantee brings unique challenges. For example, the private funders that currently back social impact bond projects are typically set up to provide loans, not letters of credit or insurance policies, as social impact guarantees require. And the insurance companies that might ultimately be best suited to offer social impact guarantee policies are largely not engaged with the Pay for Success community.

To address this challenge, I propose that social impact bond loans be used as a stepping stone. Remember, with the social impact guarantee, a service provider has to be financially able to pay back the government at some future date, should that be necessary. While it would be most natural to take out an insurance policy for that purpose, the provider could instead borrow money and immediately place it into a reserve on its balance sheet, which would then be used to pay back the lenders once impact is proven. Alternatively, if necessary, the reserve could be used to honor the government’s penalty provision.

Pencil this out and you will see that under this stepping stone approach, the lenders would experience something very much like a typical social impact bond. The money is drawn down gradually as the social services are provided, and the lenders get paid back only if the program works. Indeed, the one main difference — a simple reserve on the provider’s balance sheet — would be celebrated by the social impact bond lenders as a safer form of “counterparty risk” than the complex government appropriations promises and sinking funds commonly used in the social impact bond model.

Some readers will note that this stepping stone loan approach does not solve the double capitalization problem. However, it would retain the benefits of simplifying the government side of the contracting process. Moreover, it would set the stage for a smooth transition toward using letters of credit (which are a promise to write a check in the future if needed), rather than outright loans. Ultimately, it will be ideal to tap into the immense balance sheets of major insurance companies. I expect this will eventually happen. But in the meantime, a letter of credit approach would be a powerful step in the right direction.

There will be other challenges as well. For example, in most jurisdictions, any money returned to the government might need to be directed to a general fund, rather than to the department that originally contracted for social services. Also, the payment of social impact guarantee premiums and other aspects of setting up the Pay for Success project may fall outside the permissible use of some of the spending streams that would be tapped. These are problems that already challenge the social impact bond market. They are also limited and solvable. Although the social impact guarantee approach does not avoid them, it addresses other concerns that make it considerably less onerous than the additional challenges we have dealt with while implementing social impact bonds.

At Third Sector Capital Partners, Inc., we believe that the social impact guarantee holds more than enough potential to merit testing, as it promises to simplify the contracting process (by employing already-existing techniques), lower project costs for governments (by reducing double capitalization and the amount of time it takes to construct projects), and tap into the immense world of mainstream insurance. Ultimately, we believe it could enable Pay for Success contracting to scale more rapidly, thus empowering more governments to pursue outcomes-based contracting that measurably improves the lives of our most vulnerable citizens.

We will most certainly continue to embrace social impact bonds, as they remain the de facto vehicle for financing Pay for Success projects in the United States. That said, we plan to pursue and implement the world’s first social-impact-guarantee-financed project and grow the social impact guarantee market from there.