WhitmanTrading

What Is a Development Impact Bond?

Development impact bond is an outcomes contract, not a debt security. Investors fund a social programme up front, an outcome payer repays them with a return only if an independent evaluator verifies that agreed targets were met, and nothing is owed if they were not.

Most aid money is spent against a budget: a donor funds activities, the activities happen, and whether anything changed is a separate question asked afterwards. An impact bond inverts that order.

How it works

A price series where payment is contingent on a result.
A development impact bond pays on outcomes. Illustrative chart - not real market data.

Investors provide the working capital. Foundations, social investors or commercial funds put up the money a delivery organisation needs to run a programme, before any result exists.

A steady series with funding supplied ahead of delivery.
Investors fund the programme up front. Illustrative chart - not real market data.

A service provider runs the programme. Typically a non-governmental organisation working on something measurable — school enrolment, tuberculosis detection, employment placement.

A rising series where a payer settles against targets.
An outcome payer repays if targets are met. Illustrative chart - not real market data.

An outcome payer settles at the end. A donor agency, a foundation, or a government agrees in advance to pay a stated amount per unit of verified result, and pays nothing for effort that produced none.

A falling series where an evaluator rules on the result.
An independent evaluator decides whether they were. Illustrative chart - not real market data.

And an independent evaluator rules on it. A separate organisation measures the agreed indicators, and its finding is what releases the money.

Why it is not a bond

A choppy series with no coupon and no principal claim.
There is no coupon and no principal claim. Illustrative chart - not real market data.

A bond is a debt claim with a schedule. Interest on stated dates, principal at maturity, a ranking in insolvency, and a legal obligation that exists whether or not the borrower’s plans worked out. A corporate bond has all four.

A slow series where the instrument's name overstates its claim.
So the name bond is doing a lot of work. Illustrative chart - not real market data.

An impact bond has none of them. No coupon accrues, no principal is owed, there is nothing to rank in a liquidation because there is no borrower, and the entire return is contingent on a measurement made by a third party years later.

A calm series showing an equity-shaped payoff profile.
The risk is equity-shaped, not debt-shaped. Illustrative chart - not real market data.

The payoff shape is equity, or worse. Downside is the whole investment with no recourse; upside is capped by the contract at a fixed rate. An investor benchmarking that against fixed income is comparing it to the wrong thing entirely, which is what the name invites them to do.

A worked example

An outcome payer offers a stated amount for each additional child enrolled and retained in school. The figure is agreed before anything begins.

Investors advance the delivery budget to the organisation that will do the work, and carry the risk that the work does not produce enrolments.

A falling series with a stop level marked.
A stop fills where the market is. Illustrative chart - not real market data.

The programme runs for three years while the provider adapts its method freely, because it is being paid for a result rather than for a set of pre-approved activities.

The evaluator measures at the end. If the verified count clears the threshold, the payer settles and investors receive their capital plus the agreed return. If it falls short, they receive a reduced payment or nothing at all.

Whatever happened, the donor paid only for verified outcomes. That transfer of delivery risk from the public purse to private capital is the single thing the structure exists to achieve.

The four parties, and which one is load-bearing

The investor supplies capital and absorbs the failure case. Everything about the instrument’s risk profile follows from that one position.

The service provider does the work and gains something a restricted grant never gives it: the freedom to change method mid-programme when the method is not working.

The outcome payer supplies the money at the end, which makes its creditworthiness and its continued existence a real exposure across a multi-year term.

And the evaluator decides everything. The contract turns entirely on a measurement, so the evaluator’s independence and method are not administrative details — they are the instrument. A weak evaluation design makes the whole structure unenforceable in either direction.

How it differs from a grant

A grant pays for inputs. Money is released against a budget of activities, reported against that budget, and the provider is accountable for having spent it as promised.

An impact bond pays for a number. The provider is accountable for the indicator, and how it gets there is largely its own business.

That is a genuine improvement in one respect and a genuine hazard in another. It removes the perverse incentive to complete activities that are visibly not working, and it installs a new one to optimise for the indicator rather than the underlying goal.

The original data

This site’s thirty-year fee measurement: 5 basis points costs 1.5% of the final balance, 20 basis points costs 5.8%, 75 costs 20.2% and 150 costs 36.5%.

That table is the right lens for the transaction costs here. Structuring, legal work, contract negotiation and independent evaluation are all paid for out of the same pot as delivery, and on a small programme they are a far heavier drag than any of those fee levels.

A candlestick chart annotated with the cost of a round trip.
A round trip costs a share of a bar. Illustrative chart - not real market data.

It is also why these structures cluster at larger sizes. A fixed cost of arrangement has to be spread across enough outcome units to be worth carrying, which rules the mechanism out for exactly the small local programmes it sounds best suited to.

A price series with volume shown beneath.
Volume and price measure different things. Illustrative chart - not real market data.

What it is genuinely good for

Shifting delivery risk off the donor. A funder that pays only for verified results never funds a programme that did not work, which is not true of any grant.

Buying flexibility for the provider. Freedom to change approach part-way through is worth a great deal in fieldwork, and conventional restricted funding actively prevents it.

Forcing a measurement to exist. Agreeing the indicator, the threshold and the evaluator in advance imposes a discipline that a large share of development spending has never been subject to.

And making the price of a result explicit. Once a payer has stated what one verified outcome is worth, that figure can be compared against alternatives — which is an uncomfortable conversation, and a useful one.

When it fails

The characteristic failure is the indicator coming apart from the goal. A contract pays per child enrolled, so the provider enrols children — and the ones easiest to enrol are the ones closest to enrolling anyway, while the hardest cases, who were the reason for the programme, are quietly left out because they cost more per unit of payment. The target is hit, the evaluator verifies it honestly, the payer settles, the investors earn their return, and the underlying problem is roughly where it started. Every party behaved exactly as the contract instructed. The contract bought a number.

A candlestick series with a gap through a level.
A gap skips the level entirely. Illustrative chart - not real market data.

A second failure is attribution. Proving the programme caused the change needs a credible counterfactual, and without one the payment rewards whatever was going to happen regardless.

A third is transaction cost swallowing a share of the money that would have gone further as a grant.

A fourth is outcome-payer risk, which is ordinary credit risk sitting at the end of a multi-year term and is frequently not priced at all.

A declining series cut short at a verification decision.
Verified, but did anything change? Illustrative chart - not real market data.

And a fifth is the name. An investor who read “bond” and priced it like debt has mispriced an equity-shaped exposure, and will find that out only at the end.

Catastrophe bond covers the other instrument whose repayment hangs on an event. Corporate bond covers what an actual debt obligation looks like. And credit risk covers the exposure to the party that has to pay at the end.

What I actually do

Calling this a bond is the most consequential piece of naming in development finance. A bond is a promise to repay. This is a promise to repay if something works, which is a completely different instrument, and the name pulls in investors expecting the first thing.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.