IN ONE SENTENCE — A structurer turns a client problem that no standard product solves into a set of cash flows, legal wrappers and risk transfers that investors will buy and the bank can approve.
Where it sits. In most universal banks the structuring and solutions team sits in Global Markets, between the sales desks that face clients and the trading desks that hold risk. Some banks attach a “solutions” group to the financing side (DCM, loan capital markets) instead, and the boundary moves with each reorganisation. At Meridian the team has two halves. Cross-asset structuring builds notes, repackagings and hedges for investors and corporates. Sovereign and public-sector solutions works with finance ministries, development finance institutions (DFIs) and multilateral development banks (MDBs) on liability management, credit-enhanced financing and disaster-risk transfer. This module is about the second half, which is where debt-for-nature conversions, blended finance and catastrophe bonds are put together.
Clients. Finance ministries and debt management offices of emerging and frontier sovereigns; MDBs and DFIs that provide guarantees and concessional money; conservation organisations that design the environmental commitments and often run the conservation trust fund; insurers and reinsurers that cede catastrophe risk; and on the other side, the investors: emerging-market bond funds, insurance-linked securities (ILS) funds, pension funds with impact mandates, and impact investors.
How the desk earns money. Fees, not balance sheet. A sovereign liability management exercise pays a dealer-manager fee on the bonds tendered, usually expressed in cents or tenths of a percent of face value. The new financing pays an arrangement or structuring fee and a placement or underwriting fee. A catastrophe bond pays the structuring agent and bookrunner a fee on the notes placed. Orders of magnitude: a USD 300m to USD 1.5bn debt-for-nature conversion generates a few million dollars of fees for the arranging bank, spread over six to eighteen months of work; a mid-size cat bond produces a fee in the hundreds of thousands. The desk also creates flow for other desks: the rates or FX hedge of the new loan, the secondary trading of the new notes, the custody of the collateral. Because individual deals are rare and slow, the desk is judged on its pipeline and on the revenue it brings to the whole bank, not deal by deal.
The team. A managing director who owns the relationships with two or three MDBs and a handful of ministries; two or three structurers who build the cash-flow model and the term sheets; a quantitative analyst for pricing and risk transfer; a sovereign coverage banker who knows the country; the DCM syndicate desk that places the notes; the liability management team that runs the tender; and, on every deal, a large cast outside the desk: Legal and external counsel in two or three jurisdictions, the bank’s reputational risk function, KYC and sanctions, credit risk (for the warehouse period), product control (for any day-one profit), and the sustainable finance solutions team that checks the environmental commitments against the bank’s own framework.
A realistic week. Monday: call with the ministry of finance of a small island state on the latest version of the debt sustainability numbers. Tuesday: the structurer reruns the savings model after the MDB’s credit committee asks for a larger liquidity reserve. Wednesday: the reputational risk committee reviews the transaction for the second time, focusing on the fee table and the consultation of local fishing communities. Thursday: a call with the conservation organisation on the milestones of the marine protected areas and the penalties if they are missed. Friday: two hours with external counsel on the conditions precedent of the guarantee, because the notes cannot be priced until the guarantee is effective. Little of this is mathematics. Most of it is getting a dozen institutions to sign the same set of numbers on the same day.
WHO DOES WHAT — The structurer owns the model and the term sheet; the coverage banker owns the client; syndicate owns the pricing of the notes; liability management owns the tender; Legal owns the documents; the reputational risk committee owns the decision to put the bank’s name on the deal.
Every structured transaction answers three questions in order. Payoff: which cash flows does each party receive, in which states of the world? Wrapper: through which legal object do those cash flows reach the investor: a note issued off a bank’s or an MDB’s medium-term note programme, a note issued by a special purpose vehicle (SPV), a loan, a swap, an insurance or reinsurance contract? Documentation: which contracts make the payoff enforceable, who signs them, under which law, and which regulator looks at them?
The payoff is designed by decomposing it into building blocks that can each be priced. A principal-protected note is a zero-coupon bond plus an option. A guaranteed loan is a risky loan plus a credit default protection sold by the guarantor. A catastrophe bond is a risk-free note plus a short position in a digital or call-spread option on a natural-hazard index. Once the blocks are priced separately, the structurer knows how much “budget” there is to pay for each feature, and who should pay for it.
INTUITION — A structurer is like an architect who must use standard bricks. The client asks for an unusual building; the architect’s skill is to find the combination of standard bricks (bonds, options, guarantees, insurance) that produces the shape, and to price each brick before promising the result.
The cleanest example of a building-block payoff is an outcome bond of the kind the World Bank issued for black rhinos in South Africa in 2022 (see section 3.4). The investor buys a note at par from a AAA issuer, gives up all coupons, and receives at maturity the principal plus a “success payment” whose size depends on a measured conservation outcome. The coupons the investor gives up are not lost: the issuer redirects their value to the conservation project. A donor, not the issuer, funds the success payment.
FORMULA
is the price of a zero-coupon bond of the issuer paying 100 at maturity; the issuer’s funding rate for tenor years; the value of the coupons the investor gives up, per 100 of par, which finances the project upfront; the maximum success payment at maturity in percent of par; the achievement ratio of the outcome (0 to 100%); the investor’s annual return.
Term by term. is what the principal protection costs: with a 5-year rate of 4.30%, 81.02 per 100. , 18.98 per 100, is the project budget that the investor finances by accepting a zero coupon. The success payment is a contingent claim on the outcome, paid by a third party; if the outcome is achieved in full, the investor recovers part of the coupons forgone, never all of them in the reference deals. The investor’s concession, measured as the gap between and , is the price of impact.
INTERACTIVE LAB — Outcome note: what the investor gives up and what the project receives
Decompose a principal-protected outcome note into a zero-coupon bond and a contingent success payment, and compute the investor’s return and the funding delivered to the project.
| Parameter you can move | Range | Default |
|---|---|---|
| Issue size | 50 to 500 USD m | 150 USD m |
| Issuer funding rate for the tenor | 1 to 7 % | 4.3 % |
| Tenor | 2 to 10 years | 5 years |
| Maximum success payment at maturity | 0 to 30 % of par | 9.2 % of par |
| Outcome achieved | 0 to 100 % of target | 100 % of target |
| Result at default settings | Value |
|---|---|
| Value of principal protection (zero-coupon) | 81.02 per 100 |
| Value of coupons given up | 18.98 per 100 |
| Funding delivered to the project (present value) | 28.47 USD m |
| Success payment paid by the donor | 13.80 USD m |
| Investor return | 1.776 % |
| Investor concession against the issuer’s curve | 252.4 bp |
Try it. Set the outcome achieved to 50%: the investor still earns less than 1% a year. The concession is the price of impact, and it is paid whatever happens to the rhinos.
PITFALL — Do not describe the coupons given up as a “donation” by the investor of 4.30% a year. The project receives their present value upfront, 18.98 per 100, and the investor’s real cost is the concession on its return, which shrinks as the outcome is achieved. Both figures matter, for different readers.
When the payoff cannot be delivered by an existing issuer, the structurer creates one. A special purpose vehicle is a company with no other business, often an Irish designated activity company (DAC) or a Luxembourg securitisation vehicle, owned by a charitable trust so that it sits outside any group. It buys an asset (a loan, a bond, a portfolio) and issues notes whose cash flows mirror the asset’s, net of costs. A trustee represents the noteholders, holds the security over the asset and enforces it if needed. A corporate services provider runs the SPV’s administration; a paying agent distributes cash; an account bank holds the reserve accounts.
The point of the wrapper is to change what the investor holds. A sovereign loan is not a security that a bond fund can buy; notes issued by an SPV that owns that loan are. A loan guaranteed by an MDB carries the MDB’s credit; an SPV note backed by the guaranteed loan can be rated close to the guarantor. This is the pattern of every recent debt-for-nature conversion: a bank lends to the sovereign, the loan is sold to an SPV, the SPV issues notes, and the credit enhancement attached to the loan passes through to the notes.
Liability management means changing the composition of a borrower’s debt: buying back, exchanging or amending existing bonds. For a sovereign whose bonds trade well below par, a buyback captures the discount: every dollar of face value retired costs less than a dollar.
Mechanics. The sovereign, through a dealer-manager bank, announces a tender offer: it will buy back a series of bonds at a fixed price, or at a price set by a modified Dutch auction within a range. The offer is open for five to ten business days. Holders tender; the sovereign accepts up to a maximum amount and may set a minimum acceptance level below which it cancels the offer. The offer is conditional on the new financing closing (the “financing condition”). The dealer-manager advises on the price, canvasses investors, and is paid a fee on the face value accepted. An information and tender agent collects instructions through the clearing systems.
Pricing. Holders tender if the offer beats what they can get in the market. The offer price is therefore the pre-announcement market price plus a premium of a few points, to compensate for the fact that the announcement itself will lift the price of the bonds left outstanding. The sovereign gains if the price paid is below the value of the debt to it; we measure that value in section 5.
Exit consents are sometimes attached: tendering holders vote, as they tender, to amend the terms of the bonds left outstanding, making them less attractive and encouraging participation. They are controversial and rarely used in conservation deals, which need the goodwill of the market.
PITFALL — A buyback at 80 cents of a bond trading at 74 is not a 20% saving. The relevant comparison is with the market price (74) and with the cost of the new money that funds the buyback. If the new money costs the same as the old, the sovereign has simply paid a premium to its bondholders.
The new money is cheap only if someone stronger than the sovereign stands behind it. Four instruments are used.
The structurer’s question is how much of each the deal needs to reach the rating that the investor base requires, and at what cost.
IN ONE SENTENCE — A catastrophe bond is a note whose principal is lost, in whole or in part, if a defined natural catastrophe occurs during its life, in exchange for a high coupon.
Mechanics. A sponsor (an insurer, a reinsurer, a public pool or a government through an MDB programme) enters into a reinsurance or derivative contract with an SPV. The SPV issues notes to investors and holds the proceeds in a collateral trust invested in safe assets: money-market funds, Treasury bills, or notes of a AAA issuer such as the World Bank. Investors receive the collateral yield plus a risk spread paid by the sponsor. If a qualifying event occurs, the collateral is released to the sponsor up to the loss defined in the contract; otherwise the principal returns to investors at maturity, typically after three or four years.
Triggers. An indemnity trigger pays on the sponsor’s actual losses; an industry-loss trigger on an index of market losses; a modelled-loss trigger on the output of a catastrophe model fed with the event’s parameters; a parametric trigger on the physical parameters of the event itself, such as the central pressure of a hurricane inside a defined box, or the magnitude and depth of an earthquake. Sovereigns use parametric triggers because they pay within weeks, without loss adjustment. The price of speed is basis risk: the payout can differ from the actual damage, in both directions.
Pricing. A risk-modelling firm computes the attachment probability (that any loss reaches the notes), the exhaustion probability (that the whole principal is lost) and the expected loss. The risk spread is quoted as a multiple of expected loss; the multiple moves with the supply and demand of reinsurance capacity. The case prices one in section 5.6.
For a debt-for-nature conversion the documents fall into five sets: the tender offer memorandum and dealer-manager agreement; the loan agreement between the bank and the sovereign, and the transfer agreement to the SPV; the guarantee or insurance agreement and its conditions precedent; the offering circular of the notes, the trust deed and the agency agreement; and the conservation documents (conservation funding agreement, statutes of the conservation trust fund, milestones and penalties). A cat bond adds the reinsurance or derivative contract, the risk analysis by the modelling firm and the calculation agent agreement.
Inside the bank, a new structure goes through the new product approval process (Legal, Compliance, Risk, Operations, Finance, Tax sign off) and every transaction through the transaction approval or underwriting committee. For sovereign nature deals Meridian adds a review by the reputational risk committee (section 7).
The risks for the bank are modest in size and large in reputation: warehouse risk if the loan sits on the balance sheet before the notes are sold, execution risk if the tender fails, documentation risk across jurisdictions, and the risk of being named in a story about high fees, weak conservation or harm to local communities.
IN ONE SENTENCE — A debt-for-nature conversion refinances part of a country’s expensive debt with cheaper, credit-enhanced debt, and commits part of the saving to conservation.
The first generation, from the late 1980s, were small bilateral or NGO-funded purchases of commercial bank loans at deep discounts. The second generation, since 2021, uses the capital markets: a bank loan to the sovereign, credit-enhanced by a DFI or MDB, refinanced through SPV notes or a “blue” or “nature” bond, and a buyback of Eurobonds by tender offer.
The reference deals, as reported by Reuters in December 2024 and by a G-24 working paper of April 2026:
| Country, year | Transaction | Credit support | Conservation commitment |
|---|---|---|---|
| Belize, 2021 | Buyback of its external commercial bond at a deep discount | DFC insurance; The Nature Conservancy | USD 4m a year and a USD 23m trust |
| Barbados, 2022 | USD 150m debt conversion, 15-year blue loan | Guarantees | Protection of up to 30% of its seas |
| Ecuador, 2023 | Buyback of about USD 1.6bn face at about 40 cents | DFC USD 656m; IDB USD 85m | About USD 450m for the Galápagos |
| Gabon, 2023 | USD 500m blue bond to buy back USD 436m | DFC insurance | About USD 163m over 15 years |
| El Salvador, 2024 | USD 1bn loan | DFC USD 1bn; CAF USD 200m letter of credit | About USD 352m for the Lempa river basin |
| Bahamas, 2024 | USD 300m swap | MDB guarantee and private insurers | Mangroves and marine areas |
| Ecuador, 2024 | USD 1.53bn Amazon conversion, 17 years | DFC USD 1bn insurance; IDB | About USD 460m for the Amazon |
Barbados also closed a “debt-for-climate-resilience” conversion in 2024, a buyback of about USD 300m of domestic debt with upfront funding from the IDB and the Green Climate Fund, which redirects savings to water and food-security infrastructure. The label has widened accordingly: “debt-for-climate”, “debt-for-education” and “debt-for-development” swaps use the same engineering with a different use of the saving. Côte d’Ivoire’s 2024 operation, with a World Bank policy-based guarantee and a commitment to build schools, is the first in that family.
What changed in 2025 and 2026. Reuters reported in May 2025 that the DFC, which had insured nearly 90% of about USD 6bn of debt swapped in five years, was stepping back from climate work, putting several pipeline deals (Angola and Zambia among them) in doubt. The market has looked for substitutes: MDB guarantees, as in the Bahamas, and private insurers. In January 2026 the credit fund Enosis Capital and AXA XL announced a programme of political risk insurance for a pipeline of about USD 3bn of swaps over roughly four years. At the Fourth International Conference on Financing for Development in Seville (July 2025), Spain and the World Bank announced a Global Hub for Debt Swaps for Development to share methods and build capacity.
IN ONE SENTENCE — Blended finance uses concessional or catalytic money from public or philanthropic sources to change the risk or return of a transaction so that commercial investors will finance it.
The OECD Development Assistance Committee adopted five Blended Finance Principles in 2017; the DFI Working Group on blended concessional finance for private-sector projects applies “enhanced principles” whose core tests are additionality (the deal would not happen without the concessional money), minimum concessionality (no more subsidy than needed), commercial sustainability, market building and high standards. Convergence, the industry network that tracks the market, counted 123 blended finance transactions worth about USD 18.3bn in 2024, against a record USD 23.1bn in 2023, and warned in May 2025 that cuts in development aid budgets would weigh on future volumes.
Five instruments recur.
The economics are simple to state: the concessional money is worth the difference between the market price of the risk and the price charged. The question that matters for a commercial bank is the mobilisation ratio, private money raised per dollar of concessional money, which is also how MDBs report their own performance. A full guarantee mobilises one dollar of private capital per dollar of guarantee; a first-loss guarantee of 35% mobilises almost three. Section 5.5 computes the trade-off.
Three tools reach sovereigns.
Catastrophe bonds. The World Bank (IBRD) issues cat bonds for member countries: it is the issuer, the collateral is its own AAA notes, and the risk is transferred to investors through a contract with the country. Jamaica first obtained named-storm cover through a World Bank cat bond in 2021 and renewed it in 2024. In October 2025 Hurricane Melissa reached the pre-agreed parametric triggers, based on the storm’s central pressure and track as reported by the US National Hurricane Center, and the World Bank announced on 7 November 2025 a payout of 100% of the USD 150m bond. It was the clearest demonstration so far that a sovereign parametric cover pays quickly and in full.
The market as a whole is no longer a niche. Artemis counted a record USD 25.6bn of catastrophe bond issuance in 2025, in 122 transactions, and a record USD 61.3bn outstanding at the end of 2025. AM Best reported a record USD 17.3bn of 144A property catastrophe bonds in the first half of 2026. Most of the risk is US hurricane and earthquake risk ceded by insurers; sovereign and public-sector deals are a small share.
Regional parametric pools (the Caribbean Catastrophe Risk Insurance Facility, the African Risk Capacity, the Pacific Catastrophe Risk Insurance Company) offer annual parametric policies to member governments, reinsured in the market.
Climate-resilient debt clauses (CRDCs), or pause clauses, let a borrower defer debt service for one or two years after a qualifying disaster, with the deferred amounts capitalised and repaid later. UK Export Finance began offering them in its sovereign loans in November 2022, and a private-sector working group chaired by the UK published, through ICMA, term sheets for CRDCs in bonds and loans the same month; MDBs have since added them to their own loans. In July 2025, at the Seville conference, a Debt Suspension Clause Alliance was launched, co-led by Canada, France and the United Kingdom with the IDB, the EIB, the African Development Bank, CAF and the Asian Development Bank. The IDB said its clauses already provided USD 3.2bn of protection; the EIB had made them available to 70 developing countries.
A resilience bond links the cost of cover to investment in risk reduction: if the sponsor builds a sea wall that lowers the modelled loss, part of the premium saving is rebated. The concept is well documented; closed transactions remain few.
Social impact bonds (more accurately, outcome contracts) began with the Peterborough prison reoffending programme in the United Kingdom in 2010: private investors funded a social programme and a public payer repaid them, with a return, only if a measured outcome was achieved. Hundreds of outcome contracts have since been signed, mostly small and domestic.
Outcome bonds issued by the World Bank transfer the idea to capital markets with principal protection. The Wildlife Conservation Bond of March 2022, known as the rhino bond, raised USD 150m for five years; investors receive no coupons, the World Bank funds conservation in two South African reserves instead, and at maturity investors may receive a success payment backed by a USD 13.76m results-based grant from the Global Environment Facility, depending on the growth rate of the black rhino population as verified by independent parties. In August 2024 the World Bank priced a USD 225m, 9-year Amazon reforestation-linked bond, structured by HSBC: principal-protected, with part of the coupon forgone and redirected to a Brazilian reforestation company, and an additional coupon linked to the sale of carbon removal units generated by the project. The simulator of section 2.2 reproduces the logic of the first.
There is no single standard for these instruments. Each borrows from several.
| Instrument | Standard used | Who checks |
|---|---|---|
| Blue or nature notes | ICMA Green Bond Principles; blue economy guide (2023); nature guide (June 2025) | Second-party opinion |
| Conservation commitments | Conservation funding agreement; IFC Performance Standards for safeguards | Conservation organisation, guarantor |
| Blended finance | OECD DAC principles; DFI enhanced principles | Guarantor’s credit committee |
| Cat bonds | Risk analysis by a modelling firm; Rule 144A offering | Rating agencies, investors |
| Outcome bonds | Issuer’s sustainable development framework | Independent verifier |
On the bank’s side the regulation is general banking and markets regulation, not a sustainable-finance rule.
For investors, the notes are often bought by funds classified under Article 9 of the SFDR, whose managers need evidence that the investment contributes to an environmental objective. That demand is one reason for the second-party opinion and the reporting commitments.
WHY IT MATTERS ON THE DESK — The structurer will be asked by investors “what is green about this?” and by the reputational risk committee “who could be hurt by this?”. Both answers live in the conservation documents, which the bank does not write but must understand.
INTERACTIVE LAB — Life of a debt-for-nature conversion, from first meeting to the first conservation report
Follow the Corallia transaction through the institutions that touch it and the documents each one produces.
| # | Who | Does what | Deliverable |
|---|---|---|---|
| 1 | Minister of finance and debt management office, Corallia | Asks Meridian and two other banks for proposals to reduce debt service and fund the marine conservation plan (March 2026) | Request for proposals |
| 2 | Sovereign coverage banker and structurer, Meridian | Build a first savings model on public data, test the idea with the DFI and a conservation organisation, pass the reputational pre-screening | Proposal with indicative savings and fee table |
| 3 | Ministry of finance, Corallia | Mandates Meridian as structuring bank and dealer-manager; appoints legal counsel and a conservation partner (June 2026) | Mandate letter, engagement letters |
| 4 | Conservation partner with the ministry of environment and local communities | Draft the conservation commitments: marine protected areas, milestones, governance of the trust fund, consultation plan | Conservation plan and draft conservation funding agreement |
| 5 | DFI credit and safeguards teams | Appraise the sovereign risk, the safeguards (IFC Performance Standards) and the additionality; approve the guarantee at board level | Guarantee approval, term sheet of the guarantee |
| 6 | Meridian credit risk and reputational risk committee | Approve the warehouse limit on Corallia and the bank’s role after reviewing fees, safeguards and consultation | Credit approval, reputational risk clearance |
| 7 | Rating agencies and second-party opinion provider | Rate the guaranteed notes; review the use of proceeds and the conservation commitments | Preliminary ratings, second-party opinion |
| 8 | Liability management team, Meridian, as dealer-manager | Launch the tender offer for the 6.75% 2033 bonds, conditional on the financing (2 November 2026) | Tender offer memorandum, announcement |
| 9 | DCM syndicate, Meridian | Market and price the SPV notes with the guarantee in place (12 November 2026) | Offering circular, pricing term sheet |
| 10 | Meridian loan desk and Corallia | Sign and fund the loan, settle the tender, retire the bonds; Meridian holds the loan in warehouse | Loan agreement, tender settlement |
| 11 | Trustee, SPV and paying agent | SPV buys the loan from Meridian with the note proceeds; endowment paid to the trust fund (3 December 2026) | Transfer agreement, trust deed, closing memorandum |
| 12 | Conservation trust fund and independent verifier | Disburse the annual conservation budget; verify milestones; publish an annual report to noteholders and the DFI | Annual impact report, milestone certificate |
Origination and mandate (steps 1 to 3). The request comes from the finance ministry, often after a conversation started by a conservation organisation or an MDB. The proposal that wins is the one that shows the most credible saving with the most credible credit enhancement, not the lowest fee. Before any number goes to the client, Meridian’s sovereign coverage team runs a pre-screening with Compliance (sanctions, corruption indices, politically exposed persons) and a first reputational check: is the country in a restructuring, is the IMF programme on track, are there known conflicts over land or fishing rights?
Conservation design (step 4). This is the longest step: three to nine months. The conservation partner drafts the commitments with the environment ministry: which areas will be protected, by when, with which management plan, how much money goes into the trust fund each year, and what happens if a milestone is missed (in the Belize model, extra payments into the trust). Consultation of fishing communities and indigenous peoples is part of the safeguards and, increasingly, of the reputational test.
Guarantee approval (step 5). The DFI appraises the transaction as a sovereign operation. Its board approves the guarantee, its fee and its conditions precedent, including evidence of the conservation commitments and of the sovereign’s internal approvals (parliament, if the constitution requires it for new debt).
Bank approvals (step 6). Meridian’s credit committee approves a warehouse limit: the bank will hold the loan between its funding and its sale to the SPV. Meridian’s reputational risk committee reviews the fee table, the safeguards and the communication plan. The transaction approval committee signs off the role of dealer-manager and underwriter.
Ratings and opinion (step 7). The agencies rate the notes on the strength of the guarantee and the structure (timely payment, liquidity reserve during a claim). A second-party opinion assesses the notes against the ICMA principles and the nature guide.
Execution (steps 8 to 11). The tender, the pricing of the notes and the closing must be sequenced so that no party carries a risk it did not price. The tender is conditional on the financing; the notes are priced only when the guarantee is effective; Meridian funds the loan and settles the tender, then sells the loan to the SPV against the note proceeds. Between the two, Meridian holds a USD 315m exposure to the sovereign: the warehouse risk of section 5.8.
Life of the transaction (step 12). The SPV pays noteholders from the sovereign’s loan payments; the trust fund spends the conservation budget; a verifier certifies milestones. Meridian’s role becomes residual: it may act as calculation agent or hedge counterparty, and it remains reputationally linked to the deal.
Question. What does Corallia need, and what can a debt-for-nature conversion realistically do for it?
Data. The Republic of Corallia is a fictional small island state in the Caribbean. All data below are case assumptions.
| Item | Value |
|---|---|
| Population; nominal GDP | 280,000; USD 5.0bn |
| GDP at purchasing power parity | USD 7.5bn |
| Public debt | 88% of GDP, USD 4.4bn |
| Sovereign rating | B, stable outlook |
| Bond targeted | USD 400m 6.75% bullet maturing November 2033 (7 years) |
| Market price before announcement | 74.00 |
| Production emissions (excluding land use) | 1.8 MtCO2e a year |
| Natural asset | Coral reef, mangroves, an exclusive economic zone of about 120,000 km² |
To keep the arithmetic readable, the bond pays an annual coupon and the analysis starts on a coupon date. Market rates are also round case assumptions: a flat US Treasury curve at 4.30% and a guaranteed-note spread of 70 bp. Check live levels on the screen before reusing any of them.
Reading. Corallia pays 6.75% on a bond that the market prices to yield more than 12%. The difference between coupon and yield is the market’s view of its credit. The government’s own plan commits it to protect 30% of its exclusive economic zone by 2030, and its budget cannot pay for the patrols, the scientific monitoring and the support to fishers that this needs. The debt-for-nature conversion links the two problems: it uses the discount on the bond and a guarantee from a AAA-rated development finance institution (the DFI, a multilateral development bank in this case) to create a saving, and commits part of it to the marine plan.
Question. At what price can Corallia buy back its bonds, how much does it retire, and how large must the new loan be?
Data. Meridian’s liability management team advises a fixed-price tender at 80.00, a premium of 6 points over the pre-announcement price, and expects 90% of holders to tender. The new loan must also fund an endowment of USD 20m for the Corallia Ocean Trust and transaction costs of USD 7m (Meridian’s fees, which are detailed in section 5.8, legal counsel in three jurisdictions, rating agencies, the conservation partner, the second-party opinion, the SPV set-up).
Calculation. The pre-announcement yield is the rate that equates the bond’s remaining cash flows to its price:
is the yield to maturity. Solving gives . At the tender price of 80.00 the yield is 10.99%: holders who tender are paid as if Corallia’s credit had improved by 155 bp.
Face tendered: 90% × 400 = USD 360m. Cash paid: 360 × 80% = USD 288m. Market value of the same bonds before the announcement: 360 × 74% = USD 266.4m, so the tender premium paid to holders is USD 21.6m. New loan: 288 + 20 + 7 = USD 315m.
INTERACTIVE LAB — The buyback: tender price, face retired and size of the new loan
Price a fixed-price tender offer against the market and size the refinancing loan, including the conservation endowment and transaction costs.
| Parameter you can move | Range | Default |
|---|---|---|
| Bond outstanding | 100 to 1000 USD m | 400 USD m |
| Share of holders tendering | 50 to 100 % | 90 % |
| Coupon of the old bond | 3 to 10 % | 6.75 % |
| Remaining maturity | 2 to 15 years | 7 years |
| Market price before announcement | 40 to 100 per 100 | 74 per 100 |
| Tender price | 40 to 100 per 100 | 80 per 100 |
| Endowment of the conservation trust fund | 0 to 60 USD m | 20 USD m |
| Transaction costs funded by the loan | 0 to 20 USD m | 7 USD m |
| Nominal GDP | 1000 to 20000 USD m | 5000 USD m |
| Result at default settings | Value |
|---|---|
| Yield before announcement | 12.55 % |
| Yield at the tender price | 10.99 % |
| Face value retired | 360.0 USD m |
| Cash paid to tendering holders | 288.0 USD m |
| Market value of the same bonds before announcement | 266.4 USD m |
| Tender premium paid to holders | 21.6 USD m |
| New loan | 315.0 USD m |
| Reduction in face value of debt | 45.0 USD m |
| Reduction in face value of debt | 0.90 % of GDP |
| Annual coupon on the bonds retired | 24.30 USD m |
Try it. Raise the tender price to 85: the loan grows to USD 333m and the reduction in face value falls to USD 27m. Every point of premium is paid with borrowed money.
Result. Corallia retires USD 360m of face value and borrows USD 315m: a reduction of USD 45m, or 0.90% of GDP. The annual coupon on the bonds retired is USD 24.3m.
Reading. The reduction in face value is small: 0.9 point of a debt ratio of 88%. That is typical. Reported “debt relief” figures in this market are often larger because they add the discount captured (here 360 − 288 = USD 72m) without deducting the endowment, the costs and the fact that the new loan is still debt. The structurer should give the ministry, and the reputational risk committee, the net figure. The real gain is not in the face value but in the price of the new money, which the next stage measures.
Two execution points belong to this stage. First, the price. Before launch, the dealer-manager sounds out a few large holders under the market-sounding rules of the Market Abuse Regulation: they are wall-crossed, receive inside information, and cannot trade until the announcement. Their answers set the premium: too low and the minimum acceptance is not reached; too high and Corallia pays holders more than needed. Second, the holdouts. The 10% who do not tender keep a bond whose credit has improved, because Corallia’s debt service has fallen: their price rises after the announcement. That free ride is unavoidable in a voluntary tender and is one reason why the premium is never zero.
PITFALL — “Corallia bought back its bonds at a 20% discount” is true and misleading. The discount to par is captured only because the market prices Corallia as a weak credit; the same market would charge Corallia more than 12% for new money. Without the guarantee, buying back at 80 with money borrowed at 12.5% would create no value at all.
Question. What does the new loan cost Corallia, how much value does the credit enhancement create, and how is it shared?
Data. The loan of USD 315m is sold to an Irish SPV, Corallia Ocean Notes DAC, which funds it by issuing notes guaranteed by the DFI. Terms (case assumptions): tenor 15 years, five years of grace then ten equal annual repayments (weighted average life 10.5 years); notes priced at 4.30% + 70 bp = 5.00%; guarantee fee 1.00% a year on the outstanding amount; SPV running costs (trustee, agents, administration, ratings surveillance) 0.10% a year. Corallia pays the sum, 6.10%, on the loan. In the conservation funding agreement Corallia commits USD 3.0m a year for 15 years to the trust fund, on top of the USD 20m endowment.
The value of debt to Corallia is measured by discounting its debt service at Corallia’s own market cost of borrowing, 12.55%, the yield of section 5.2. This is the convention of most structurers’ models; the pitfall below discusses it.
FORMULA
is the value created by the credit enhancement; the loan principal (USD 315m); and the interest and repayment paid in year on the new loan; the discount rate, here Corallia’s market yield. If Corallia could borrow only at , would be zero: the loan would be worth its face. is the gift of the cheaper rate.
Calculation. Computed with the note rate plus SPV costs (5.10%), before the guarantee fee, the gross value is = USD 129.8m. With the guarantee fee (6.10%), the present value of the debt service is USD 202.6m and = 315 − 202.6 = USD 112.4m. The difference, USD 17.4m, is the present value of the guarantee fees paid to the DFI. The rest is shared out as follows.
INTERACTIVE LAB — Who captures the value of the credit enhancement
Compute the present value created by the guaranteed refinancing and split it between guarantor, bondholders, fees, conservation and the sovereign’s budget, at a discount rate of your choice.
| Parameter you can move | Range | Default |
|---|---|---|
| Face value retired | 100 to 1000 USD m | 360 USD m |
| Coupon of the old bond | 3 to 10 % | 6.75 % |
| Remaining maturity of the old bond | 2 to 15 years | 7 years |
| Tender price | 40 to 100 per 100 | 80 per 100 |
| Endowment of the trust fund | 0 to 60 USD m | 20 USD m |
| Transaction costs | 0 to 20 USD m | 7 USD m |
| Yield of the guaranteed notes | 3 to 8 % | 5 % |
| Guarantee fee | 0 to 3 % a year | 1 % a year |
| SPV running costs | 0 to 0.5 % a year | 0.1 % a year |
| Tenor of the new loan | 6 to 20 years | 15 years |
| Grace period | 0 to 8 years | 5 years |
| Annual conservation payment | 0 to 10 USD m | 3 USD m |
| Discount rate (Corallia’s market yield) | 3 to 20 % | 12.55 % |
| Result at default settings | Value |
|---|---|
| New loan | 315.0 USD m |
| All-in rate paid by Corallia | 6.10 % |
| Value of the old debt service retired | 266.3 USD m |
| Value of the new debt service | 202.6 USD m |
| Value created by the enhancement, before guarantee fee | 129.8 USD m |
| Share of the guarantor (fees) | 17.4 USD m |
| Value created, after guarantee fee | 112.4 USD m |
| Share of bondholders | 21.7 USD m |
| Share of conservation (endowment and annual payments) | 39.8 USD m |
| Share of Corallia’s budget | 43.9 USD m |
| Conservation as share of conservation plus budget | 47.6 % |
| Annual cash saving in the grace period, before conservation | 5.085 USD m |
| Annual cash saving in the grace period, after conservation | 2.085 USD m |
| Annual repayment after the grace period | 31.50 USD m |
| Price of the old bond at the discount rate | 73.99 per 100 |
Try it. Lower the discount rate to 6.10%, the all-in rate of the new loan: the value created by the enhancement falls to zero, yet Corallia still gains USD 29.1m, because at that rate the bonds it bought at 80 were worth 103.6.
Result. Of USD 129.8m of gross value, the guarantor takes USD 17.4m, bondholders USD 21.7m, fees and costs USD 7.0m, conservation USD 39.8m and Corallia’s budget USD 43.9m, the largest single share. Conservation receives 47.6% of what goes to Corallia and nature together. In cash terms during the grace period, Corallia pays USD 19.2m of interest instead of USD 24.3m of coupon: a saving of USD 5.085m a year, of which USD 3.0m goes to the trust fund and USD 2.085m stays in the budget.
Reading. Three points for the ministry and for the credit committee.
First, the saving exists only because of the guarantee: it is a transfer of the DFI’s credit standing to Corallia, priced at 1.00% a year. The DFI captures USD 17.4m of the value; the rest is the subsidy implicit in a fee well below the market price of Corallia’s risk (section 5.4).
Second, conservation is financed by the sovereign’s budget, not by the bondholders or the bank. The trust fund receives value that would otherwise have stayed with the Treasury. Whether that is a good use of fiscal space is a political choice; a structurer should not present it as free money for nature.
Third, the cash relief in the first five years is real but small next to Corallia’s budget: USD 2.085m a year net. The larger effect is on the maturity profile: USD 360m due in 2033 is replaced by repayments of USD 31.5m a year from 2032 to 2041, at a fixed rate.
PITFALL — The result depends on the discount rate. At Corallia’s market yield of 12.55% the gain looks large because any cheap money looks valuable. At 8%, a rate closer to what the IMF uses in some debt sustainability work, Corallia’s gain is USD 36.5m and the “bondholder share” becomes negative: at that rate the bonds bought at 80 were worth 93.5. State the rate, show two of them, and never compare savings across deals computed at different rates.
Question. Is a 1.00% fee a fair price for guaranteeing a B-rated sovereign for 15 years, and what does the guarantee do to Corallia’s cost of funds?
Data. One-year probability of default for a B credit: 4.00% (the course scale). If the guarantee is called, the DFI pays investors and becomes a creditor of Corallia. Because sovereigns have historically kept paying MDBs through crises, the DFI expects to recover most of what it pays: its loss given a call is assumed at 15%, against about 50% for commercial creditors in a sovereign restructuring (case assumptions).
Calculation. Expected loss of the DFI per year: 4.00% × 15% = 0.60% of the amount guaranteed, or USD 1.89m on USD 315m. Fee income: USD 3.15m. Margin over expected loss: USD 1.26m a year, before the DFI’s own costs and the cost of the capital it sets aside. The fee would just cover expected loss if the loss given a call were 1.00% / 4.00% = 25%. Over 15 years, a 4% annual PD implies a cumulative probability of default of 45.8%: the DFI is not insuring a remote event.
On the other side, Corallia’s market spread is 12.55% − 4.30% = 825 bp. The guaranteed structure costs it 70 + 100 + 10 = 180 bp over Treasuries. The guarantee compresses Corallia’s spread by 645 bp.
INTERACTIVE LAB — The guarantor’s view: fee against expected loss
Compare the DFI’s guarantee fee with its expected loss for a range of loss-given-call assumptions, and measure the spread compression that the guarantee delivers to the sovereign.
| Parameter you can move | Range | Default |
|---|---|---|
| Amount guaranteed | 100 to 1500 USD m | 315 USD m |
| Guarantee fee | 0 to 3 % a year | 1 % a year |
| One-year PD of the sovereign | 0.1 to 15 % | 4 % |
| DFI loss given a call | 0 to 60 % | 15 % |
| Tenor | 5 to 20 years | 15 years |
| Sovereign market yield | 4 to 20 % | 12.55 % |
| Treasury yield | 1 to 7 % | 4.3 % |
| Spread of the guaranteed notes | 20 to 200 bp | 70 bp |
| SPV running costs | 0 to 0.5 % a year | 0.1 % a year |
| Result at default settings | Value |
|---|---|
| DFI expected loss | 0.60 % a year |
| Fee income, first year | 3.15 USD m |
| Expected loss, first year | 1.89 USD m |
| Fee minus expected loss | 1.26 USD m |
| Loss given call at which fee equals expected loss | 25.0 % |
| Cumulative probability of default over the tenor | 45.8 % |
| Sovereign market spread | 825 bp |
| All-in spread of the guaranteed structure | 180 bp |
| Spread compression delivered by the guarantee | 645 bp |
Try it. Set the loss given a call to 50%, the haircut of a commercial creditor: the expected loss is twice the fee. The deal works only because the DFI’s preferred treatment is assumed to hold.
Result. At the assumed 15% loss given a call, the fee covers expected loss with a margin of USD 1.26m a year. At 25% it covers nothing more; at 50% the DFI would lose USD 3.15m a year in expectation.
Reading. The whole structure rests on one assumption: that a sovereign which might restructure its bonds will keep paying the DFI. History supports it on average, which is why rating agencies and MDBs give weight to preferred creditor treatment, but it is a practice, not a legal priority. Two consequences for the desk. The fee is not a market price of Corallia’s risk; it is a policy price, set by the DFI’s board, and the subsidy it contains is what makes the conversion possible. And the DFI’s capacity to give such guarantees is limited by its own capital and by its exposure limits per country; a DFI that guarantees USD 315m for Corallia cannot use that headroom for a loan to Corallia’s water utility. The additionality test (section 7) asks whether this is the best use of that headroom.
Question. The DFI’s credit committee asks whether a guarantee of the first 35% of losses, instead of 100% of payments, would do the job and free capacity for other operations. What happens to the notes and to mobilisation?
Data. Same loan of USD 315m, same 4.00% PD. In a sovereign default the loan is assumed to suffer a haircut of 50% in the restructuring; in a severe case 70%. Under a first-loss guarantee, the DFI pays investors their realised loss up to 35% of the principal, and nothing above. The loss it pays is final: the restructuring has already reduced the claim, so there is nothing further to recover, unlike the full guarantee where the DFI steps into the investors’ claim.
FORMULA
is the amount guaranteed; the first-loss share (35%); the loan; the haircut in a default; the loss borne by noteholders as a share of the loan; their expected loss per year; the mobilisation ratio, private financing per dollar of guarantee.
Calculation. Guarantee amount: 35% × 315 = USD 110.25m. Mobilisation: 315 / 110.25 = 2.86, against 1.0 for the full guarantee. In a default with a 50% haircut, noteholders lose 50% − 35% = 15% of principal, USD 47.25m; their expected loss is 4.00% × 15% = 0.60% a year. With a 70% haircut they lose 35%. To protect noteholders against the severe case the first-loss share must reach 70%, which brings the guarantee to USD 220.5m and the mobilisation ratio down to 1.43.
INTERACTIVE LAB — First-loss guarantee: protection of the notes against mobilisation of private money
See how the size of a first-loss guarantee drives the noteholders’ expected loss and the private money mobilised per dollar of guarantee, for several restructuring haircuts.
| Parameter you can move | Range | Default |
|---|---|---|
| Loan funded by the notes | 100 to 1500 USD m | 315 USD m |
| One-year PD of the sovereign | 0.1 to 15 % | 4 % |
| Haircut on the loan in a default | 10 to 90 % | 50 % |
| First-loss share guaranteed by the DFI | 5 to 100 % | 35 % |
| Result at default settings | Value |
|---|---|
| Amount guaranteed | 110.25 USD m |
| Private financing per dollar of guarantee | 2.86 x |
| Noteholders’ loss in a default | 47.25 USD m |
| Noteholders’ expected loss | 0.60 % a year |
| DFI payment in a default | 110.25 USD m |
Try it. Move the first-loss share from 35% to 50% with a 50% haircut: the notes become loss-free in the base case, while the DFI still mobilises two dollars per dollar of guarantee.
Result. A 35% first-loss guarantee mobilises 2.86 dollars per dollar of guarantee but leaves the notes with an expected loss of 0.60% a year in the base case, comparable to a sub-investment-grade credit. A full guarantee makes the notes close to risk-free and mobilises 1.0.
Reading. This is the central trade-off of blended finance. A first-loss structure stretches scarce concessional capacity further, but the notes then need investors who accept emerging-market credit risk, at a spread well above 70 bp, and the saving for Corallia shrinks. A full guarantee reaches the deepest pool of investors at the lowest spread, but consumes the guarantor’s capacity one for one. The DFI also changes its own risk: under the first-loss structure it pays the haircut, a loss that its preferred treatment no longer protects, so each dollar it guarantees is lost in full in a default: its expected loss per dollar guaranteed equals the PD, 4.00% a year, against 0.60% under the full guarantee. The DFI chose the full guarantee for Corallia. A structurer should present both options with numbers; MDB boards increasingly ask for the mobilisation ratio of every operation.
Question. A hurricane could wipe out the fiscal gain in a single season. What do a pause clause and a parametric catastrophe bond add, and at what cost?
Data. The loan agreement includes a climate-resilient debt clause: if a hurricane meets a parametric trigger, Corallia may defer up to two years of debt service; deferred amounts are capitalised at the loan rate and repaid over the remaining life. Separately, the ministry considers a three-year parametric cat bond issued through an MDB programme: cover USD 100m, attachment probability 4.0% a year (a 1-in-25-year storm), exhaustion probability 2.0% (1-in-50), risk multiple 2.0, collateral yield 3.90% (all case assumptions).
Calculation, pause clause. Debt service in the grace period is 315 × 6.10% = USD 19.215m a year. A two-year deferral keeps USD 38.43m in the Treasury at the moment it is needed most. Capitalised at 6.10%, the deferred amounts become USD 39.6m at the end of the deferral. For investors the deferral is close to neutral in present value at the loan rate, and the guarantor covers the payment risk in the meantime, which is why the DFI accepted the clause.
Calculation, cat bond. With losses spread evenly between attachment and exhaustion, the expected loss of the layer is approximately the average of the two probabilities:
is the annual probability that the event reaches the attachment point; the probability that it exhausts the cover; the risk multiple demanded by investors; the risk spread paid by the sponsor. Investors receive the 3.90% collateral yield plus 6.0%, a coupon of 9.90%. Corallia pays USD 6.0m a year for USD 100m of cover, 2.88 times its net annual cash saving from the conversion.
INTERACTIVE LAB — Resilience layer: catastrophe bond premium and value of a debt-service pause
Price a parametric catastrophe bond from attachment and exhaustion probabilities, compare its premium with the conversion’s annual saving, and compute the liquidity a pause clause provides.
| Parameter you can move | Range | Default |
|---|---|---|
| Cat bond cover | 25 to 300 USD m | 100 USD m |
| Attachment probability | 0.5 to 10 % a year | 4 % a year |
| Exhaustion probability | 0.25 to 8 % a year | 2 % a year |
| Risk multiple of expected loss | 1 to 5 x | 2 x |
| Collateral yield | 0 to 6 % | 3.9 % |
| Net annual saving of the conversion | 0.5 to 20 USD m | 2.085 USD m |
| Loan subject to the pause clause | 100 to 1500 USD m | 315 USD m |
| Loan rate | 2 to 10 % | 6.1 % |
| Years of debt service deferred | 1 to 3 years | 2 years |
| Result at default settings | Value |
|---|---|
| Expected loss of the layer | 3.00 % a year |
| Risk spread | 6.00 % a year |
| Investor coupon | 9.90 % |
| Annual premium paid by Corallia | 6.00 USD m |
| Return period of the attachment event | 25 years |
| Premium as a multiple of the net annual saving | 2.88 x |
| Annual debt service in the grace period | 19.215 USD m |
| Liquidity kept by the pause clause | 38.43 USD m |
| Deferred amount capitalised at the loan rate | 39.60 USD m |
Try it. Lower the attachment probability to 2% (a 1-in-50-year storm) with exhaustion at 1%: the premium halves, but the cover pays in fewer of the hurricanes that would actually hurt Corallia’s budget.
Result. The pause clause gives USD 38.43m of liquidity at almost no cost; the cat bond gives USD 100m of cover for USD 6.0m a year.
Reading. The two tools are complements. The pause clause protects the debt service of this one loan; the cat bond pays cash for reconstruction, within weeks, as Jamaica’s did after Hurricane Melissa in 2025. But the premium is almost three times the net saving of the conversion: Corallia cannot pay for meaningful cover out of the conversion alone. In practice premiums for small states are often supported by donors, which is another form of blended finance. The structurer’s honest message is that a debt-for-nature conversion improves the debt profile at the margin; it does not by itself make a hurricane-exposed island resilient. Basis risk must also be explained to the ministry in plain words: a storm that passes just outside the box, or with a central pressure just above the threshold, pays nothing.
Question. What exactly does the money buy for nature, and how is delivery enforced?
Data. The conservation funding agreement, signed by Corallia, the conservation partner and the trust fund, and referred to in the loan agreement, contains four commitments (case assumptions modelled on the public features of the reference deals):
| Commitment | Deadline | Evidence |
|---|---|---|
| Adopt a marine spatial plan for the whole EEZ | End of 2028 | Law published, plan filed with the trust |
| Designate 30% of the EEZ as marine protected areas, a third of it no-take | End of 2030 | Legal designation, maps |
| Fund the trust: USD 20m endowment and USD 3.0m a year | From closing | Payments recorded by the trust |
| Management plans and enforcement for each area | Two years after designation | Verifier’s report |
If a milestone is missed, the annual payment to the trust rises by USD 0.5m for each year of delay; the extra money stays in conservation. A missed milestone is not a default under the loan: the noteholders and the DFI are paid regardless.
Reading. Four design choices deserve attention, because they are where critics look first.
The commitments are not credit terms. Making a conservation milestone an event of default would make the notes unrateable at the guarantor’s level and would punish the country twice in a crisis. The consequence is that enforcement relies on the penalty payments, the governance of the trust and reputation. Investors who buy the notes as “blue” are buying a credible plan, not a contractual guarantee of outcomes.
Governance of the trust fund. The trust is an independent foundation with a board where the government holds a minority of seats, alongside the conservation partner, local fishers’ cooperatives and a scientific member. It publishes audited accounts. In the reference deals, critics have pointed to the limited participation of local communities and to the oversight shifted from parliament to offshore vehicles. Corallia’s structure gives the fishers a seat and requires public consultation on each management plan, at a cost of time: the conservation design took seven months.
Additionality of the conservation spending. Some of what the trust funds may already have been in the budget. The conservation partner documents a baseline: Corallia spent about USD 0.8m a year on marine management before the deal (case assumption). The agreement requires the budget line to be maintained, so that the trust’s money adds to it rather than replacing it.
Measurement. The milestones measure legal acts and management inputs, not ecological outcomes such as reef cover or fish biomass, which respond slowly and are hard to attribute. The trust commissions a scientific baseline in 2027 and repeats it every five years, so that outcomes are at least observed, even if they do not trigger payments.
WHY IT MATTERS ON THE DESK — When the reputational risk committee asks “what if Corallia ignores the commitments?”, the structurer must answer precisely: the bondholders are paid, the DFI is paid, the trust receives penalty payments, and the reputational exposure falls on everyone whose name is on the offering circular, including Meridian’s.
Question. What does Meridian earn, what capital does it use, and what is the risk that matters?
Data. Meridian’s fees (case assumptions, disclosed to the ministry in the fee table): structuring and arrangement fee 0.60% of the loan; dealer-manager fee 0.25% of the face value tendered; placement fee 0.35% of the notes. Cost-income ratio 60% on fee revenues (course convention). Operational-risk capital: under the CRR3 standardised approach, own funds are a percentage of the business indicator, 15% at the margin for a bank of Meridian’s size, and the EU has set the internal loss multiplier to 1; Meridian allocates to each deal RWA of 12.5 × 15% = 1.875 times its revenue. Warehouse: Meridian funds the loan on 12 November and sells it to the SPV on 3 December, 21 days. If the guarantee is effective during the warehouse, the exposure is weighted at 0% by substitution (the DFI is a multilateral development bank listed in CRR Article 117(2)); if not, Meridian holds an unguaranteed exposure to a B-rated sovereign, weighted 100% under the standardised approach, with PD 4.00% and LGD 45%.
Calculation.
| Line | Computation | USD m |
|---|---|---|
| Arrangement fee | 315 × 0.60% | 1.890 |
| Dealer-manager fee | 360 × 0.25% | 0.900 |
| Placement fee | 315 × 0.35% | 1.103 |
| Revenues | 3.893 | |
| Operating costs | 60% | -2.336 |
| Expected loss (guarantee effective) | 0 | |
| Pre-tax result | 1.557 | |
| Net result after 25% tax | 1.168 | |
| RWA: operational risk | 1.875 × 3.893 | 7.298 |
| Allocated capital | 13% × 7.298 | 0.949 |
Meridian’s fees are 124 bp of the loan, a little over half of the USD 7m of transaction costs.
INTERACTIVE LAB — Meridian’s RAROC on the mandate: fees, operational-risk capital and warehouse
Rebuild the RAROC of a structuring mandate line by line and see how a warehouse without effective guarantee, or a delayed placement, consumes the return.
| Parameter you can move | Range | Default |
|---|---|---|
| Loan arranged and placed | 100 to 1500 USD m | 315 USD m |
| Face value tendered | 100 to 1500 USD m | 360 USD m |
| Arrangement fee | 0 to 1.5 % of loan | 0.6 % of loan |
| Dealer-manager fee | 0 to 1 % of face tendered | 0.25 % of face tendered |
| Placement fee | 0 to 1 % of notes | 0.35 % of notes |
| Cost-income ratio on fees | 30 to 90 % | 60 % |
| Marginal operational-risk coefficient | 12 to 18 % | 15 % |
| Days in warehouse | 0 to 180 days | 21 days |
| Status of the guarantee during the warehouse | Guarantee effective / Guarantee not yet effective | Guarantee effective |
| Risk weight of the unguaranteed sovereign | 0 to 150 % | 100 % |
| PD of the sovereign | 0.1 to 15 % | 4 % |
| LGD of the sovereign loan | 10 to 75 % | 45 % |
| Result at default settings | Value |
|---|---|
| Fee revenues | 3.893 USD m |
| Operating costs | 2.336 USD m |
| Expected loss in warehouse | 0.000 USD m |
| Net result after tax | 1.168 USD m |
| Operational-risk RWA | 7.298 USD m |
| Warehouse RWA, time-weighted | 0.000 USD m |
| Allocated capital (13% of RWA) | 0.949 USD m |
| RAROC after tax | 123.1 % |
| Meridian’s fees on the loan | 123.6 bp |
| Days of unguaranteed warehouse at which RAROC falls to 12% | 42 days |
Try it. Choose ‘Guarantee not yet effective’ and move the warehouse from 21 to 60 days: a fee business with a three-digit RAROC falls below the hurdle because the bank sat on a B-rated sovereign it never meant to hold.
Result. With the guarantee effective during the warehouse, RAROC is 123.1%: a fee business with almost no balance sheet. If the guarantee is not yet effective when Meridian funds the loan, the 21-day warehouse adds USD 18.1m of time-weighted RWA and USD 0.33m of expected loss, and RAROC falls to 27.9%. It falls below the 12% hurdle after 42 days and becomes negative if placement is delayed by four months.
Reading. Capital is not the constraint in structuring; sequencing is. The most important line of the closing checklist is “guarantee effective before funding”, and the credit committee approves the warehouse limit on that condition. The second constraint, which does not appear in RAROC, is reputational: a fee of USD 3.9m on a conservation deal is defensible if it is disclosed, comparable to a plain liability management exercise of the same size and complexity, and small next to the value created for Corallia. It is indefensible if it is hidden.
| Stage | Key figure | Value |
|---|---|---|
| 5.2 | Yield before announcement; at tender price | 12.55%; 10.99% |
| 5.2 | Face retired; cash paid; tender premium | USD 360m; 288m; 21.6m |
| 5.2 | New loan; reduction in face value | USD 315m; USD 45m (0.90% of GDP) |
| 5.3 | All-in rate of the new loan | 6.10% |
| 5.3 | Value created, gross; after guarantee fee | USD 129.8m; USD 112.4m |
| 5.3 | Guarantor; bondholders; fees | USD 17.4m; 21.7m; 7.0m |
| 5.3 | Conservation; Corallia’s budget | USD 39.8m; USD 43.9m |
| 5.3 | Annual saving in grace, before and after conservation | USD 5.085m; USD 2.085m |
| 5.4 | DFI expected loss; breakeven loss given call | 0.60% a year; 25% |
| 5.4 | Spread compression | 825 bp to 180 bp, 645 bp |
| 5.5 | First-loss 35%: guarantee; mobilisation; notes EL | USD 110.25m; 2.86x; 0.60% |
| 5.6 | Pause clause liquidity; cat bond premium | USD 38.43m; USD 6.0m a year |
| 5.8 | Meridian revenues; RAROC; with unguaranteed warehouse | USD 3.893m; 123.1%; 27.9% |
Read the table from the bottom up and the case tells one story. The bank earns a modest fee at very little capital, provided it never holds the sovereign unguaranteed. The guarantor lends its credit at a policy price. The sovereign gains a better debt profile, not solvency. Nature receives a share of the sovereign’s gain, under commitments that are enforced by money and reputation rather than by default clauses.
IN ONE SENTENCE — A structuring mandate consumes operational-risk capital on its fees and, for a few weeks, credit-risk capital on the warehouse; the guarantee decides whether the second is zero or large.
Credit risk on the warehouse. Meridian treats the exposure to Corallia under the standardised approach (a case assumption; many IRB banks keep sovereign exposures there under the permanent partial use provisions). Under CRR Article 114, a central government with a B rating (credit quality step 5) is weighted 100%; below B−, 150%. The DFI guarantee is recognised as unfunded credit protection if it meets the CRR requirements: a direct claim on an eligible protection provider, explicit, irrevocable and unconditional, enforceable in all relevant jurisdictions, and covering timely payment. Multilateral development banks are eligible providers, and those listed in Article 117(2) are weighted 0%. By substitution, the guaranteed part of the exposure takes the guarantor’s weight: 0%. The same substitution applies in the large exposures regime, so the warehouse counts against Meridian’s limit on the DFI, not on Corallia, provided the guarantee is effective at funding.
Operational risk. CRR3 replaced the previous approaches with a single standardised approach: own funds equal the business indicator component, which applies marginal coefficients of 12%, 15% and 18% to successive buckets of the business indicator (a three-year average of income components), and the EU has set the internal loss multiplier to 1. Meridian allocates operational-risk RWA to each fee-generating deal at 12.5 × 15% = 1.875 times its revenue, a simple internal convention.
What is not there. No market-risk capital, since Meridian keeps none of the notes; no counterparty credit risk, since there is no hedge in the base case; no securitisation framework, since the SPV notes are backed by a single loan and the guarantee is not tranched (the first-loss variant of section 5.5 would raise the question whether a tranched guarantee creates a securitisation position for investors, which counsel must answer).
The table of section 5.8 is the credit and transaction application. Two observations.
First, with only operational-risk capital, RAROC does not depend on the size of the fees: it is for any fee amount. Fee businesses are judged on absolute contribution and on the time of scarce people, not on RAROC. The figure that changes the decision is the warehouse: with the guarantee not yet effective RAROC falls to 27.9% for 21 days and below the hurdle after 42 days (simulator of section 5.8).
Second, what does the “nature” feature change? Not the price: the dealer-manager, arrangement and placement fees are those of a plain liability management exercise and a guaranteed financing of the same size. It changes three things. Feasibility: without the conservation commitments there is no DFI guarantee, and without the guarantee Corallia cannot refinance at 6.10%; a plain tender funded by a new Eurobond at 12.5% would create no value. Cost: the conservation design, the second-party opinion and the extended reputational review lengthen the mandate from three months to nine, a cost borne in the 60% cost-income ratio only on average; Meridian’s structuring team records about 1,200 hours on the deal (case assumption). Risk: the reputational exposure is larger than for a plain liability management exercise, and it lasts for the 15 years of the notes.
Treasury charges the warehouse at the short-term USD funding cost plus a liquidity premium for the actual holding period, not for the 15-year contractual maturity of the loan, on one condition: the sale to the SPV is committed at funding (a forward purchase agreement with the SPV, conditional only on note settlement). Without that commitment Treasury would charge a 15-year term premium, about 90 bp on the course’s scale of 6 bp per year of tenor, and the NSFR would require stable funding for a long-dated loan to a non-investment-grade sovereign at the 85% factor applicable to performing loans with a risk weight above 35%. With the sale committed, the warehouse is a short-term asset in Treasury’s view. The underwriting commitment on the notes is also reported as an off-balance-sheet item in liquidity monitoring until the notes settle.
Fees (IFRS 15). The dealer-manager, arrangement and placement fees are recognised when each performance obligation is satisfied, in practice at settlement of the tender and at closing; fees contingent on closing are not recognised before. Costs incurred on an abandoned mandate are expensed.
Warehouse loan (IFRS 9). The loan is originated in order to be sold: its business model is neither “hold to collect” nor “hold to collect and sell”, so it is measured at fair value through profit or loss. Any difference between the funding amount and the price paid by the SPV is a trading result. On sale, the loan is derecognised because Meridian transfers substantially all risks and rewards and keeps no continuing involvement.
SPV (IFRS 10). The SPV is owned by a charitable trust; Meridian, as arranger, has no exposure to its variable returns after closing and no power over its relevant activities, so it does not consolidate it. If Meridian held notes, acted as swap counterparty or had decision rights, the analysis would have to be redone.
If Meridian had kept part of the loan. The DFI guarantee would be integral to the contractual terms and taken into account in measuring expected credit losses; the stage would follow Corallia’s credit risk, but the loss given default would reflect the guarantee.
GAR and BTAR. Exposures to central governments are excluded from both the numerator and the denominator of the Green Asset Ratio, so neither the warehouse nor any retained piece would move it (see A02).
Pillar 3 ESG. Under the EBA templates, banking-book exposures that support climate change mitigation or adaptation without being Taxonomy-aligned can be reported in the template for other climate change mitigating actions. Meridian holds nothing after closing, so the deal appears in no Pillar 3 template; it may appear in the narrative on sustainable finance activity.
Financed and facilitated emissions (PCAF). The PCAF standard for facilitated emissions (Part B, December 2023) excludes sovereign debt and use-of-proceeds instruments from its scope, so the arranging role produces no facilitated emissions figure. Financed emissions (Part A) attribute to a lender of a sovereign a share of the country’s emissions equal to its exposure divided by GDP at purchasing power parity. Had the warehouse straddled Meridian’s reporting date, the attribution would have been material.
INTERACTIVE LAB — Financed emissions of a sovereign exposure (PCAF Part A)
Attribute a share of a country’s production emissions to a lender in proportion to its exposure over PPP-adjusted GDP.
| Parameter you can move | Range | Default |
|---|---|---|
| Exposure to the sovereign at reporting date | 0 to 1000 USD m | 315 USD m |
| GDP at purchasing power parity | 1000 to 50000 USD m | 7500 USD m |
| Production emissions, excluding land use | 0.1 to 20 MtCO2e | 1.8 MtCO2e |
| Result at default settings | Value |
|---|---|
| Attribution factor | 4.20 % |
| Financed emissions | 75.6 ktCO2e |
| Financed emissions per USD m lent | 240 tCO2e |
Try it. Set the exposure to zero, the position after the sale to the SPV: Meridian reports nothing, although it built the whole transaction.
The attribution factor would have been 4.20%, and Meridian would have reported 75.6 ktCO2e of financed emissions for an exposure held three weeks. The asymmetry of the standards (large figures for a short hold, nothing for the arranging role) is one reason why banks also report a “sustainable finance mobilised” metric that includes arranged deals. The trap is double counting: the DFI reports the same USD 315m as private finance mobilised, the conservation partner as conservation finance unlocked, and each investor as a blue investment. Meridian’s methodology should say which share of the deal it counts and why.
| Risk | What can go wrong | Control | Owner |
|---|---|---|---|
| Warehouse credit risk | Loan funded before the guarantee is effective; placement delayed | Condition precedent “guarantee effective”; committed SPV purchase; warehouse limit | Credit risk, structurer |
| Execution risk | Tender undersubscribed; financing condition fails | Minimum acceptance level; pre-sounding of holders under wall-crossing rules | Liability management |
| Documentation risk | Guarantee not enforceable or not eligible as credit protection | Legal opinions in each jurisdiction; CRM eligibility review | Legal, Risk |
| Fee transparency | Fees and costs not disclosed to the ministry or the public | Full fee table in the proposal and in the offering circular | Coverage, Compliance |
| Conflict of interest | Bank advises on the tender price while holding the bonds | Information barriers; disclosure of holdings; no proprietary tender | Compliance |
| Safeguards | Restrictions on fishing without consultation; harm to communities | DFI safeguard review; consultation plan; grievance mechanism | Conservation partner, DFI |
| Basis risk | Parametric cover does not pay after a damaging storm | Basis-risk analysis explained to the ministry in writing | ILS structurer |
| Reputational risk | Deal described as “relief” that is mainly a subsidy to bondholders | Net figures (section 5.3) in all communications; committee review | Reputational risk committee |
| Corruption and sanctions | Payments diverted through the trust or the SPV | KYC on all parties, PEP screening of the trust board, audited accounts | Compliance |
Red-flag checklist. Escalate to the reputational risk committee if any of the following is true.
PITFALL — The worst greenwashing in this business is not a false environmental claim; it is a true environmental claim attached to a misleading financial one. A marine reserve can be real and the “USD 1bn of debt relief” still be mostly a subsidy to bondholders and arrangers.
Debt-for-nature and debt-for-climate conversions.
Blended finance. Convergence counted 123 transactions worth about USD 18.3bn in 2024, against USD 23.1bn in 2023, and warned of the effect of aid cuts (Devex, May 2025). The UK-based IIED estimated that the 49 most debt-vulnerable poorer countries could swap a quarter of the more than USD 430bn they owe, an upper bound rather than a forecast (cited by Reuters, May 2025).
Catastrophe bonds. Record issuance of USD 25.6bn in 2025 across 122 transactions and a record USD 61.3bn outstanding at the end of 2025 (Artemis, January 2026); a record USD 17.3bn of 144A property cat bonds in the first half of 2026 (AM Best, September 2026). Jamaica’s USD 150m World Bank cat bond paid out in full after Hurricane Melissa (World Bank, 7 November 2025).
Pause clauses. Debt Suspension Clause Alliance launched in Seville on 1 July 2025; the IDB reported USD 3.2bn of protection under its clauses and the EIB had offered them to 70 countries.
Outcome bonds. World Bank rhino bond, USD 150m, 5 years, March 2022, with a USD 13.76m GEF results-based grant behind the success payment; World Bank Amazon reforestation-linked bond, USD 225m, 9 years, August 2024, its largest outcome bond so far.
Open debates.
INTERACTIVE LAB — Structuring and solutions: check your understanding
Fourteen questions on mechanics, value split, credit enhancement, resilience instruments and the bank’s treatment.
Q1. In the Corallia case, why does buying back bonds at 80 create value for the country?
Answer: B. Without the guarantee, Corallia would refinance at about 12.5% and the buyback would create no value; the enhancement creates USD 129.8m gross.
Q2. What is the reduction in the face value of Corallia’s debt?
Answer: C. USD 360m is retired and USD 315m is borrowed. USD 72m is the discount captured, before endowment and costs.
Q3. Who funds the conservation commitments, economically?
Answer: D. The trust receives part of the value that would otherwise go to the budget; the bondholders are paid a premium.
Q4. Lowering the discount rate used to value Corallia’s debt from 12.55% to 8% does what to the bondholders’ share?
Answer: A. At 8% the old bond is worth 93.49, above the tender price of 80.
Q5. Why can a full guarantee priced at 1.00% a year be acceptable to the DFI on a B-rated sovereign with a 4% PD?
Answer: B. With a 15% loss given a call, expected loss is 0.60% a year; at 25% the fee only covers expected loss.
Q6. A 35% first-loss guarantee on a USD 315m loan mobilises how much private financing per dollar of guarantee?
Answer: C. 315 / 110.25 = 2.86, against 1.0 for a full guarantee.
Q7. What is the main drawback of the first-loss structure for the notes?
Answer: B. Noteholders lose 15% of principal in a default with a 50% haircut.
Q8. Which trigger lets a sovereign receive a cat bond payout within weeks, without loss adjustment?
Answer: C. Parametric triggers use physical parameters of the event; the price is basis risk.
Q9. A cat bond layer has an attachment probability of 4% and an exhaustion probability of 2%. With a risk multiple of 2, the risk spread is approximately:
Answer: C. Expected loss is about (4% + 2%) / 2 = 3%; times 2 gives 6%.
Q10. What does a climate-resilient debt clause do?
Answer: B. The deferral is close to neutral in present value for lenders and gives liquidity to the borrower.
Q11. Why is the RAROC of Meridian’s mandate the same (123.1%) whatever the size of the fees?
Answer: B. (1 − 60%) × (1 − 25%) / (13% × 1.875) = 123.1%.
Q12. During the warehouse, the guarantee is not yet effective. Under the standardised approach, the B-rated sovereign is weighted:
Answer: C. Credit quality step 5 for central governments is weighted 100%; with an effective MDB guarantee, substitution gives 0%.
Q13. How does the Corallia warehouse affect Meridian’s Green Asset Ratio?
Answer: C. Sovereign exposures are outside the GAR altogether.
Q14. Under PCAF Part B (December 2023), what facilitated emissions does Meridian report for arranging the Corallia notes?
Answer: C. Part B excludes sovereign debt and use-of-proceeds instruments.
The Republic of Isla Verde has a USD 600m 5.5% bullet bond with 6 years to maturity (annual coupon), trading at 68. A tender at 73 is expected to attract 85% of holders. The new loan must also fund a USD 15m trust endowment and USD 9m of costs. Compute the yields at 68 and 73, the face retired, the cash paid, the tender premium, the new loan and the reduction in face value. The new loan costs 5.60% all-in, interest-only for the first years: what is the annual cash saving on interest?
Solution. The yield at 68 is 13.65%; at 73, 12.08%. Face retired: 85% × 600 = USD 510m. Cash paid: 510 × 73% = USD 372.3m. Tender premium over the market: 510 × 5% = USD 25.5m. New loan: 372.3 + 15 + 9 = USD 396.3m. Reduction in face value: 510 − 396.3 = USD 113.7m. Coupon on the bonds retired: 510 × 5.5% = USD 28.05m; interest on the new loan: 396.3 × 5.60% = USD 22.19m; saving USD 5.86m a year before any conservation payment. The deep discount (68) explains why the face reduction is larger than in the Corallia case.
A DFI considers guaranteeing a USD 500m loan to a sovereign with a one-year PD of 3%. Option A: full guarantee at 0.80% a year; the DFI expects to lose 20% of what it pays. Option B: guarantee of the first 40% of losses; the haircut in a default is assumed at 60%. Compute, for A, the DFI’s expected loss, its margin and the breakeven loss given a call; for B, the amount guaranteed, the mobilisation ratio and the noteholders’ loss and expected loss.
Solution. A: expected loss 3% × 20% = 0.60% a year, USD 3.0m; fee USD 4.0m; margin USD 1.0m a year. Breakeven loss given call: 0.80% / 3% = 26.7%. B: guarantee 40% × 500 = USD 200m; mobilisation 500 / 200 = 2.5. In a default noteholders lose 60% − 40% = 20% of principal, USD 100m; expected loss 3% × 20% = 0.60% a year. Option B frees USD 300m of guarantee capacity but delivers notes with a sub-investment-grade expected loss, so the spread, and the country’s saving, will be worse.
Solution. (a) Expected loss ≈ (3% + 1%) / 2 = 2.0%; risk spread 2.5 × 2.0% = 5.0%; premium USD 7.5m a year. (b) Revenues: 2.5 + 1.8 + 2.0 = USD 6.3m; costs USD 3.78m; operational-risk RWA USD 11.81m; capital USD 1.54m; RAROC = 2.52 × 0.75 / 1.536 = 123.1%. Warehouse: time-weighted RWA 500 × 30 / 365 = USD 41.10m; expected loss 500 × 4% × 45% × 30 / 365 = USD 0.74m; total RWA USD 52.91m, capital USD 6.88m; RAROC = (6.3 − 3.78 − 0.74) × 0.75 / 6.88 = 19.4%. Still above the hurdle, but the warehouse has cost five-sixths of the return.
1. Walk me through a debt-for-nature swap. A sovereign whose bonds trade at a discount buys them back through a tender offer, funded by a new loan from a bank. The loan is credit-enhanced by a DFI guarantee or political risk insurance and sold to an SPV that issues highly rated notes, so the sovereign pays a much lower rate than its market yield. Part of the saving funds conservation through an endowment and annual payments to a trust fund, under a conservation agreement with milestones. The bank earns dealer-manager, arrangement and placement fees.
2. Where does the saving really come from? From the credit enhancement. Buying back at a discount is only valuable if the new money is cheaper than the market’s price of the sovereign’s risk. In the Corallia case the guarantee compresses the spread from 825 bp to 180 bp; without it, the buyback would create no value. The discount to par is a consequence of the weak credit, not a source of gain on its own.
3. How would you split the value of the transaction between the parties? Compute the value created as the loan principal minus the present value of the new debt service at the sovereign’s market yield, before and after the guarantee fee. Then allocate: the guarantor’s fees, the bondholders’ premium over the value of their bonds, the transaction costs, the conservation endowment and payments, and the residual to the budget. Show it at two discount rates, because the answer depends on that choice.
4. Full guarantee or first-loss? A full guarantee gives the notes the guarantor’s credit and the lowest spread, but uses the guarantor’s capacity one for one. A first-loss guarantee mobilises more private money per dollar but leaves noteholders with sovereign risk above the attachment, so the spread and the country’s saving worsen. The choice depends on the guarantor’s capacity, the investor base and the country’s need; MDBs increasingly ask for the mobilisation ratio.
5. What is basis risk in a parametric cat bond and how do you explain it to a minister? It is the gap between what the bond pays and the actual damage. A storm can devastate the island while passing just outside the trigger box, and pay nothing; or a storm can trigger a payout with modest damage. I would show the minister historical storms replayed through the trigger, including the near misses, and put it in writing before signing.
6. How does the bank’s capital treat this deal? Fees attract operational-risk capital under the CRR3 standardised approach, proportional to revenue. The warehouse loan attracts credit-risk capital: 100% risk weight for a B-rated sovereign under the standardised approach, or 0% by substitution if an eligible MDB guarantee is effective at funding. That is why “guarantee effective before funding” is a condition of the credit approval.
7. What would make you refuse to put the bank’s name on such a deal? Undisclosed or disproportionate fees, a “relief” figure that the numbers do not support, commitments that only relabel existing spending, no real consultation of affected communities, or a country heading for a restructuring that the swap would complicate. Each is on our red-flag list and goes to the reputational risk committee.
8. Why are outcome bonds principal-protected, and what does the investor really pay? Because the investor base is mainstream fixed-income buyers who cannot take project risk on principal, the issuer is AAA and repays par regardless. The investor pays through the coupons it forgoes, whose present value funds the project upfront; a donor funds a success payment that partly compensates if the outcome is achieved. In the stylised rhino-type note of section 2.2, the investor earns about 1.78% at full success against 4.30% on a plain bond, a concession of about 252 bp.
Standards and market guidance
Regulation
Transactions, data and market reports
Theory