---
slug: social-impact-bond
type: pattern
summary: "An outcomes-based contract that lets a government or other payer pay only after independently verified results, with private or philanthropic investors funding delivery up front."
created: 2026-05-06
updated: 2026-06-06
related:
  change-theory:
    relation: depends-on
    note: A social impact bond needs a stated causal pathway before the parties can choose outcomes, prices, and evaluation rules.
  independent-verification:
    relation: depends-on
    note: Outcome payment depends on an evaluator whose findings trigger repayment rather than on self-reported provider activity.
  additionality-test:
    relation: tested-by
    note: The additionality question asks whether the SIB structure changed service delivery, capital availability, or payer behavior relative to ordinary contracting.
  blended-finance-stack:
    relation: complements
    note: Some SIBs sit inside broader blended structures with philanthropic guarantees, first-loss layers, or technical-assistance money.
  catalytic-firstloss-capital:
    relation: complements
    note: A family foundation or DAF sponsor can provide a first-loss layer when the SIB needs risk absorption to attract senior investors.
  impact-theater:
    relation: prevents
    note: Clear outcome pricing, independent evaluation, and loss-bearing investor capital keep the structure from becoming a public-relations contract.
---

# Social Impact Bond

> **Pattern**
>
> A named solution to a recurring problem.

*An outcomes-based contract that lets a government or other payer pay only after independently verified results, while private or philanthropic investors fund delivery up front and bear the risk that the results do not arrive.*

*Also known as: pay-for-success bond, pay-for-success financing, social outcomes contract, development impact bond, social benefit bond.*

Do not read the word *bond* too literally. A social impact bond is closer to a milestone contract than to a tradable security: one party pays for verified results, another funds the work before those results exist, and investors accept loss if the outcomes miss. The structure is useful only when the parties can write the outcome, data source, comparison method, payment cap, and loss rule before anyone signs.

## Context

A social impact bond is not a bond in the ordinary fixed-income sense. It is an outcomes contract with investor financing attached. The outcome payer, usually a government agency and sometimes a foundation or donor collaborative, agrees to pay only if a defined social outcome is achieved. Investors provide working capital to an intermediary or service provider before the outcome is known.

An independent evaluator measures the result. If the target is met, the payer repays principal and a return according to the contract. If the target is missed, investors lose some or all of their capital.

The first widely cited SIB launched in the United Kingdom in 2010 around the One Service at HMP Peterborough, where investor capital funded support for short-sentenced prisoners after release. The reported final result was a 9.0% reduction in reoffending against the comparison group, above the 7.5% threshold for repayment. That origin story explains why the instrument still appears in family-office, philanthropy, and government-innovation conversations. It promised to join prevention, private capital, public savings, and impact measurement in one structure.

The promise is real, but narrow. SIBs work best where the outcome is specific, the measurement window is short, the intervention has evidence behind it, administrative data are available, and the payer has a defensible reason to pay for prevention after the fact. They disappoint when the outcome is broad, slow, politically contested, or too expensive to evaluate cleanly. A family office should treat the instrument as a specialty contract, not as a general badge for impact seriousness.

## Problem

Many social programs face the same financing problem: the payer would rather pay for verified outcomes than fund untested activity, but the service provider can't deliver the work without money up front. Government procurement often pays for inputs or services. Philanthropy can fund pilots, but it may not want to carry a program indefinitely. Private investors may be interested in prevention economics, but they need a repayment rule.

The SIB answers that gap by moving performance risk to investors. That makes the structure attractive to an outcome payer that wants to pay for results, but it also makes the structure expensive. The parties need lawyers, evaluators, data-sharing agreements, governance rules, intervention-management capacity, and a payment formula that can survive public scrutiny.

The trap is prestige. A SIB can become a conference-stage artifact: complex enough to impress, small enough not to change a system, and expensive enough that transaction costs consume the learning. The structure earns its place only when the outcome, price, counterfactual, and data can be stated before closing.

## Forces

- **Payment discipline versus transaction cost.** Paying only for verified outcomes can protect public or philanthropic money, but the contract may cost too much to build for a small program.
- **Investor risk versus service-provider stability.** Investors should bear performance risk, while providers still need enough operating certainty to serve people well.
- **Measurement rigor versus human complexity.** Reoffending, homelessness, employment, health, and education outcomes need real evidence, but people don't live inside clean evaluation windows.
- **Innovation versus proven intervention.** The structure is often sold as innovative, yet it works best when the service model already has evidence and the financing is the uncertain piece.
- **Public savings versus public value.** Some outcomes produce budget savings. Others are worth paying for even when savings are indirect, delayed, or spread across agencies.

## Solution

Use a social impact bond only when five conditions hold: a named outcome payer, a proven or plausibly proven intervention, a measurable outcome, a credible counterfactual, and investors willing to lose money if the outcome is not achieved.

Start with the payer's duty. The contract should say who pays, why that payer is authorized to pay, which population is covered, what outcome triggers payment, and how much the payer will pay for each unit of verified result. "Reduced recidivism" is not enough. The agreement needs the cohort, baseline, comparison method, measurement date, threshold, maximum payment, and audit rights.

Then separate the parties. A clean SIB usually has five roles: the outcome payer, the service provider, an intermediary or special-purpose vehicle, investors, and an independent evaluator. The intermediary manages contracts, capital calls, provider payments, performance reporting, and investor communication. The evaluator does not manage delivery. If the evaluator also designs the intervention, the office should ask whether independence is real.

Price the outcome against value, not enthusiasm. A payer may be willing to pay $8,000 for a verified stable-housing outcome because avoided shelter, emergency, and jail costs make that price defensible. A foundation may pay because the outcome advances mission, even if fiscal savings are split across agencies. Either way, the memo should show the payment cap and investor return. If no one can explain why the outcome is worth the price, don't finance the structure.

Finally, write the failure case into the documents. A SIB is credible because investors can lose money. The Rikers Island project in New York is the canonical caution: it was the first U.S. SIB, backed by a Goldman Sachs loan and a Bloomberg Philanthropies guarantee, and it ended after the evaluation found no measurable reduction in recidivism. That result was disappointing, but it also showed the instrument's discipline. The outcome did not arrive. The full success payment did not happen.

> **⚠️ Contested question**
>
> The field still argues over whether SIBs are worth their transaction costs. Treat the structure as appropriate only when outcome pricing, evaluation design, and data access are strong enough to make the contract enforceable. A weak SIB is often worse than an ordinary grant or service contract because it adds complexity without improving accountability.

## How It Plays Out

Consider a $1.1B single-family office with a $120M foundation and a family council mandate around youth employment in the city where the family company operated for three generations. The city wants to reduce repeat unemployment among 18-to-24-year-olds leaving the juvenile justice system, but it doesn't want to expand a program line until results are visible. A nonprofit provider has a coaching and employer-placement model with prior evidence, but it needs working capital for a three-year cohort.

The proposed SIB finances services for 1,200 participants. The city is the outcome payer. The family foundation provides a $2M first-loss note. Two family-office co-investors provide $6M of senior notes. An intermediary holds the contracts and pays the service provider quarterly. A university evaluator measures whether participants are employed for at least six months within twelve months of enrollment, compared with a matched administrative-data baseline.

The payment formula is capped at $10.4M:

| Role | Commitment | Contract job | Risk position |
|---|---:|---|---|
| City outcome payer | Up to $10.4M | Pays only for verified employment outcomes above threshold. | No payment for missed outcomes. |
| Family foundation | $2M first-loss note | Absorbs initial losses and makes the senior notes investable. | First dollars at risk. |
| Co-investor senior notes | $6M | Fund service delivery and evaluation costs. | Repaid after outcomes payments, above the first-loss layer. |
| Service provider | Contracted delivery budget | Provides coaching, placement, and retention support. | Performance pressure but no investor downside. |
| Independent evaluator | Fixed evaluation contract | Determines whether payment threshold is met. | Must stay separate from delivery management. |

If verified employment outcomes exceed the threshold, the city pays enough for the foundation to recover principal and for senior investors to earn a modest return. If outcomes land just above break-even, the senior notes are repaid and the foundation may recover only part of its note. If outcomes miss the threshold, the foundation loses first, senior investors take remaining losses, and the city does not pay for a result it did not receive.

The investment committee approves the structure because the file answers the hard questions before closing. The theory of change links coaching, employer matching, and retention support to six-month employment. The evaluator has access to wage and justice-system data. The family foundation's first-loss layer changes the senior investors' willingness to participate. The public report can claim only the verified outcome and the foundation's financing role, not every life consequence the provider hopes will follow.

A weaker version would fail quietly. The city might choose a broad "improved life outcomes" target, let the provider self-report surveys, and set a payment cap that no one can tie to public value. The family office would still get an elegant diagram. It wouldn't get an underwritable SIB.

## Consequences

**Benefits.** The structure forces the parties to define success before money moves. The payer sees the outcome price. The investor sees the risk. The provider receives working capital. The evaluator has a rule to apply. A family office can participate without pretending its capital is a grant, a PRI, and a market-rate investment at the same time.

SIBs also sharpen additionality claims. The office can ask whether its capital changed the payer's willingness to contract, the provider's ability to scale, or the investor group's willingness to bear performance risk. If the answer is no, the deal may still be useful, but the SIB label is doing more reputational work than financing work.

**Liabilities.** The structure is slow and expensive. It can push providers into managing toward measured targets while missing unmeasured needs. It can exclude smaller community organizations that can't absorb reporting load. It can also make public officials appear innovative while shifting difficult social-risk questions into a contract no voter will read.

The mature use is restrained. Use SIBs for bounded outcomes with strong data, strong providers, and a payer that has a real reason to pay after verification. Use simpler grants, PRIs, or service contracts when the problem is too early, too relational, too long-horizon, or too hard to reduce to a payment trigger.

## Sources

- Social Finance, [*Reducing Reoffending in Peterborough*](https://www.socialfinance.org.uk/people/what-we-do/reducing-reoffending-in-peterborough), current access 2026 — Social Finance's account of the 2010 Peterborough SIB, including the One Service structure and the reported 9.0% reduction in reoffending.
- MDRC, [*What We Learned From the Nation's First Social Impact Bond*](https://www.mdrc.org/node/16173), 2015 — the Rikers Island post-mortem, including the no-effect evaluation result, investor loss, and lessons on evidence, data, and program fit.
- Urban Institute Pay for Success Initiative, [*What Is Pay for Success?*](https://pfs.urban.org/discover/content/what-pay-success-pfs.html), current access 2026 — a practitioner explanation of PFS roles, outcome payment, investor risk transfer, and independent evaluation.
- Akinchan Jain, World Bank, [*Five Ways for Social Impact Bonds to Live Up to Their Potential*](https://blogs.worldbank.org/en/voices/five-ways-social-impact-bonds-live-their-potential), 2019 — a critical review of SIB limitations around risk pricing, issuer risk retention, scale, transaction costs, and data.

---

*This entry describes a structural pattern and is not legal, tax, or investment advice. Consult qualified counsel and tax advisors licensed in your jurisdiction before adopting any structure described here.*

---

- [Next: Outcomes Fund](outcomes-fund.md)
- [Previous: Donor-Advised Fund as Patient Capital](daf-patient-capital.md)