Why Resilience Bonds Could Not Fund Prevention

For years, resilience bonds have been presented as a promising solution to a persistent problem in climate finance: how to pay for prevention.

The logic behind them is intuitive. If investments in flood protection, wildfire mitigation, or grid hardening reduce future losses, then those avoided losses should have economic value. And if that value can be estimated, it should be possible to finance prevention upfront and repay investors over time.

On paper, it makes sense.
In practice, it hasn’t worked.

Understanding why resilience bonds failed to fund prevention at scale requires looking beyond the instrument itself and examining the assumptions it relied on: assumptions inherited from insurance and capital markets that prevention does not fit neatly into.

Prevention is defined by absence

The core challenge is deceptively simple: prevention succeeds when nothing happens.

A levee that holds.
A fire that never spreads.
A heat wave that does not overwhelm the grid.

The value created by prevention is real, but it is counterfactual. It exists in the gap between what did happen and what would likely have happened otherwise. That makes prevention fundamentally different from the outcomes financial markets are accustomed to pricing.

Markets are very good at pricing events. They are much worse at pricing non-events.

Resilience bonds attempted to convert this counterfactual value into a contractual cash flow. That required estimating how much loss was avoided, translating that estimate into insurance savings, and then treating those projected savings as a reliable source of repayment.

This is where the structure began to strain.

Counterfactual value is hard to contract on

Avoided loss can be modeled, but it cannot be observed directly. Models evolve. Assumptions are revised. New data arrives. Two credible analysts can reasonably disagree about how much risk reduction a given project actually delivered.

Resilience bonds asked insurers or public entities to commit, in advance, to recognizing those model-based estimates as enforceable payment obligations. That is a difficult commitment to make, especially in systems already under financial and regulatory pressure.

If repayment depends on agreeing that a specific amount of loss was avoided, then every model update becomes a potential dispute. Over time, that uncertainty makes the instrument fragile.

The problem was not that the modeling was unsophisticated. It was that the model was doing too much work.

Incentives did not align

Resilience bonds also assumed incentive alignment that rarely exists in practice.

Insurance contracts are short-term. They are repriced annually and can be withdrawn when risk becomes unattractive. Prevention investments, by contrast, deliver benefits over decades and across entire regions.

Those benefits are often shared by multiple insurers, future policyholders, public agencies, and taxpayers. In many cases, any single insurer would capture only a fraction of the value created by prevention while bearing the full upfront cost.

From a private balance sheet perspective, that tradeoff is difficult to justify, even if the broader system benefits.

Resilience bonds implicitly assumed insurers would fund long-term, system-wide risk reduction in exchange for uncertain future savings. That assumption proved optimistic.

Complexity became a liability

As these challenges became apparent, resilience bond structures grew increasingly complex. They required baseline risk models, counterfactual scenarios, premium projections, regulatory approvals, and long-term agreements among multiple stakeholders.

Each additional layer introduced uncertainty. Each uncertainty reduced investor confidence.

What might have worked as a bespoke pilot struggled to become repeatable infrastructure. Capital markets, which depend on clarity and enforceability, largely stepped back.

The deeper issue: prevention was forced into the wrong frame

At a deeper level, resilience bonds failed because they attempted to force prevention into a risk-transfer framework.

They borrowed the logic of catastrophe bonds (triggers, payouts, monetized loss) and tried to invert it. But prevention does not trigger. It does not pay out. It changes probabilities over time.

By treating prevention as a derivative of insurance savings, resilience bonds remained tethered to the very architecture that struggles most to finance prevention in the first place.

The issue was not a lack of creativity or intent. It was a mismatch of architecture.

What this reveals

Resilience bonds did not fail because prevention lacks value. They failed because the financial system lacked a way to recognize that value without forcing it through instruments designed for loss.

This distinction matters.

Most institutions understand that prevention is cheaper than recovery. The problem is not awareness. It is structure. We have built sophisticated systems for pricing damage, but very little infrastructure for financing its absence.

Until that changes, prevention will remain widely acknowledged, and persistently underfunded.

A necessary shift

As climate volatility increases, there will come a point where risk transfer alone is no longer sufficient. When insurance capacity withdraws and losses migrate onto public balance sheets, the question will no longer be whether prevention is worthwhile, but how it can be financed at scale.

Resilience bonds were an early attempt to answer that question. Their limitations are instructive.

They show that prevention cannot simply be bolted onto existing risk-transfer instruments. It requires its own financial architecture—one that accepts uncertainty, aligns payers with long-term exposure, and does not depend on proving exactly what did not happen.

That is the direction climate finance will eventually have to move.

The only open question is how long we wait before redesigning the system to get there.

Where to go from here

What resilience bonds ultimately reveal is not a lack of interest in prevention, but a lack of financial architecture designed for it. If prevention is to move from being conceptually appealing to operationally fundable, it cannot rely on instruments built for loss and transfer. It requires structures that begin from different assumptions: that benefits are probabilistic, that payers are tied to long-term exposure rather than short-term contracts, and that success does not need to be proven through a single avoided event.

I am exploring what that architecture could look like through ongoing work at Arctica Risk, where the focus is on financing risk reduction as a first-class objective rather than a derivative of insurance. You can learn more at ArcticaRisk.com.