Multilateral Insurance

01 The economics of risk cover

A dollar of well-placed risk cover moves more capital than a dollar of lending, because it changes the risk arithmetic for every other investor at the table rather than just adding one more participant. The coalition’s insurance function is built entirely on that arithmetic: absorb precisely the risk the market cannot yet price — political transition, transfer restriction, first-of-kind technology, unproven counterparties — and private capital will carry everything else at its own price.

The current book mobilises $4.60 of external capital per dollar of cover outstanding, against $3.20 for the portfolio as a whole — the leverage premium of insurance over lending, visible in one comparison. The corollary is discipline: cover misplaced is leverage in reverse, amplifying losses the same way it amplifies mobilisation, which is why this page spends as much time on what the coalition refuses to cover as on what it offers.

One framing helps outsiders read everything below: the function is run as an insurer, not as a guarantee-stamping desk. Expected-loss pricing, reserves, exposure limits, published claims — the machinery is actuarial. The development mandate decides which risks are worth holding; the insurance discipline decides at what price and how much.

02 The three instrument families

Cover is engineered per commitment from a defined set — the instrument follows the risk, exactly as lending instruments follow the constraint. Each family answers a different question a financier is asking.

Partial guarantees dominate the book because they are the sharpest tool: they move exactly one named risk off a lender’s table and nothing else, which keeps the lender honest on every risk it retains. A guarantee that covers everything teaches the lender to underwrite nothing — the moral-hazard failure that sank whole guarantee programmes in earlier decades, and the reason “partial” is a design principle here rather than a budget constraint.

First-loss layers do their work in structured transactions: by absorbing the first defined tranche of losses, they transform the risk mathematics of every senior position above them, often converting an unrateable structure into an investment-grade one. The coalition sizes first-loss with actuarial care — thick enough to matter, thin enough that the seniors still feel their own risk.

Pooled cover is where multilateral scale becomes product. Weather, commodity, and disaster risks that would overwhelm any single-economy book become diversified, priceable exposures across three regions — and where reinsurance markets will take the pooled risk efficiently, the coalition cedes it, converting its origination advantage into capacity without hoarding the exposure.

• Partial guarantees — cover a named risk slice (payment default, transfer & convertibility, breach of concession) for a defined period, leaving all other risk with the lender
• First-loss layers — reorder who absorbs the first shock, so senior capital can price to the layer above it
• Pooled cover — correlated exposures (weather, commodity, disaster) pooled across the portfolio and, where efficient, transferred to reinsurance markets

03 Why multilateral pooling works

Risk that is uninsurable for one economy becomes manageable across twenty — the oldest theorem in insurance, applied where it has historically been applied least. A drought that would break a single agricultural book is a priced event across three regions; a political transition that freezes one market is noise in a pooled one.

The coalition’s scale is therefore the product itself: it can hold and diversify exposures no single member could carry, and offer cover at prices a standalone insurer of frontier risk never could. There is also a subtler advantage — information. Underwriting political and transfer risk requires knowing the politics, and the coalition’s member-owned structure gives it standing information no commercial underwriter can buy: the finance ministry whose risk is being priced sits on the owner’s side of the table, with disclosure obligations.

Membership changes incentives too. A government weighing an action that would trigger coalition cover — a convertibility freeze, a concession breach — is weighing an action against a pool it co-owns and votes in. The deterrent is quiet and unmeasurable and, on the claims record to date, apparently real.

04 What is covered — and what is not

Discipline defines the product as much as appetite does. The coverage lists are published, and the refusals are the load-bearing half.

The last exclusion is the one that surprises newcomers: cover passes the same four underwriting tests as capital. A transaction that fails additionality cannot buy a guarantee instead — insurance is not the discount door into the coalition. And the integrity carve-out is absolute: a counterparty whose own corruption causes the loss has no claim, which is written into every contract and has been enforced.

• Covered — transfer & convertibility, political violence and expropriation, regulatory breach by a public counterparty, first-of-kind technology performance, defined weather and disaster triggers
• Not covered — ordinary commercial risk a market can price, currency speculation, losses arising from a counterparty’s own integrity failures, and any exposure the underwriting tests would decline as a loan

05 Pricing and reserves

Cover is never free comfort. Every instrument is priced from expected loss plus a capital charge, reserves are held against the modelled book, and total exposure is disclosed in the coalition’s risk statements alongside the lending portfolio. Where concessionality exists in a premium, it is explicit, measured, and justified by development return in the pricing memo — exactly as with lending.

The reserve methodology is deliberately conservative: stress scenarios are drawn from each economy’s own worst historical decade, not from a benign global average, and correlated-loss scenarios assume the pooling breaks down partially in a systemic event — because in the events that matter, it partially does. Callable capital stands behind the reserves, sized so that the modelled one-in-two-hundred-year book survives without a call.

The cumulative claims ratio below one percent reflects three things in honest proportion: disciplined selection, the deterrent effect of coalition presence in a transaction, and a young book that has not yet lived through its full cycle. The coalition prices as though the third factor dominates, which is why the reserves look excessive against the record — they are meant to.

06 Claims, honestly

An insurer that never pays is a fraud; one that pays carelessly is insolvent. The claims philosophy threads that needle procedurally: triggers are pre-agreed and documented at signing — not renegotiated in the crisis — and a valid claim is paid on the contractual timeline without diplomatic weather delays. Counterparties learn quickly which promises are real, and the product’s entire value rests on this one being.

Every claim, paid or denied, is reviewed by the Risk & Audit Committee and disclosed with its rationale. The record to date is short and instructive: few claims, all resolved within contractual timelines, two of which produced amendments to standard trigger language now used across the book. A claims record treated as a learning corpus rather than an embarrassment is, in this business, the difference between an insurer and a slogan.

07 Reading the exhibits

The exhibits below show the book’s composition and its leverage. Partial guarantees carry the bulk of outstanding cover, consistent with their precision; and the mobilisation chart explains the whole strategy in one ranking — transfer-and-convertibility cover moves the most external capital per dollar, because it is the single risk international lenders most systematically refuse to hold, and therefore the exact place a multilateral balance sheet adds the most value.

08 How cover is originated and approved

Cover follows the same path as capital, with one addition. Requests originate with the transaction — a lender seeking a transfer-risk wrap, a structure needing a first-loss layer — through the regional secretariat handling the underlying commitment. The insurance desk then runs its own underwriting on top of the deal team’s: risk decomposition, trigger drafting, expected-loss modelling against the economy’s history, and a reserves impact assessment.

The addition is the exposure gate. Beyond the four tests and portfolio limits that all commitments face, cover passes an insurance-book review: aggregate exposure by risk class, correlation with existing cover, and reinsurance capacity where the pooled book would otherwise concentrate. A guarantee can clear its transaction perfectly and still be trimmed because the book already holds too much of that weather, that corridor, or that counterparty class. Cover decisions publish with the commitment, including price basis — concessionality in a premium is disclosed exactly as it is in a loan.

09 A worked example: covering a transfer-risk tranche

A $140 million renewables portfolio in a member economy stalled at the last hurdle: two European institutional lenders would take construction risk, operating risk, and counterparty risk — but not the risk that a future currency regime would trap their debt service onshore. Classic unpriceable residue: real, historically grounded, impossible for an outside lender to model.

The coalition wrote a transfer-and-convertibility guarantee on the debt-service stream alone — not the principal, not the operating risk, just the named slice — for a premium priced off the economy’s own convertibility history plus the capital charge. The lenders closed at a spread reflecting the risks they kept. Coalition exposure: a modelled expected loss a fraction of the premium stream. Mobilisation on the cover dollar: above seven to one, well over the book average, because the instrument was aimed at the exact bottleneck.

The deterrent coda: the member’s central bank sits in the coalition’s governance, and the guarantee documentation gives the coalition standing consultation rights on exchange-regime changes affecting covered transactions. The risk did not just get priced; it got a seat at the table where it would be decided. That is the part of the product no commercial insurer can copy.

10 Reinsurance and market development

The book cedes into commercial reinsurance markets wherever they will take the risk efficiently — pooled weather and disaster cover especially — for two reasons beyond capacity. Ceding disciplines pricing: if professional reinsurers will not touch a layer at any sensible price, that is information the coalition’s own models must answer to. And ceding builds the market: every treaty teaches reinsurers a frontier risk class they had refused from lack of data, using the coalition’s loss history as the syllabus.

The long game is explicit and published in the insurance strategy: risks the coalition holds exclusively today should migrate — first to shared treaties, then to commercial markets entirely, with the coalition stepping back to the next unpriceable frontier. Transfer risk in the coalition’s most established member economies is already partially syndicated. The function measures itself by the same exit test as everything else here: cover the market takes over is cover that worked.

11 Governance of the insurance book

Risk cover concentrates the coalition’s hardest governance questions — contingent liabilities are where institutions quietly overextend — so the book carries its own oversight layer. The insurance desk reports monthly to Risk & Compliance and quarterly to the Risk & Audit Committee, which reviews exposure by risk class, reserve adequacy, and every claim without management present at least once a year. Aggregate cover outstanding, reserves, and the claims register publish in the annual risk disclosures at the same standard as the lending book.

Two structural safeguards complete the frame. The desk cannot underwrite its own limits: exposure caps by risk class and economy are set by the committee, and a cover proposal that would breach them dies procedurally, exactly as a loan would. And the callable-capital backstop is stress-published: the disclosures show, in numbers, how far the modelled book can deteriorate before a call — a figure the coalition prefers its members, rating counterparties, and critics to compute from the same page it does.