Public Doctrine Series

A Contract on a Measurement

Jamaica’s catastrophe bonds after Beryl and Melissa, and what a parametric trigger is for

GFE-PUB-DOC-20261017-A · 17 October 2026
Global Internal Governance Chamber · Global Financial Engineering, Inc.
Published jointly by Global Financial Engineering, Inc. and Global Accountancy Institute, Inc.

In July 2024, days after Hurricane Beryl passed south of Jamaica as a major hurricane and left heavy damage behind it, Global Accountancy Institute published two essays on the island’s catastrophe bond. The bond, issued by the World Bank in May 2024 and known as IBRD CAR Jamaica 2024, provided $150 million of protection against named storms over four hurricane seasons. Beryl did not trigger it. The essays concluded that the trigger was badly designed: a single measurement, the storm’s minimum central pressure within a set of boxes drawn around the island, could not capture what a hurricane actually does. They proposed adding sustained wind speed, storm surge and rainfall as triggers, enlarging or overlapping the boxes, and paying more as more thresholds were crossed.

Fifteen months later the same bond was tested by a far greater storm. Hurricane Melissa made landfall in Jamaica on 28 October 2025 at Category 5, with winds reported at 185 miles an hour and a minimum central pressure of 892 millibars — the strongest landfall in the island’s recorded history. On 7 November the World Bank confirmed that the calculation agent had found the bond’s parametric conditions met in full. Jamaica received the entire $150 million within weeks. In May 2026 it returned to the market for a third bond, kept the same basic design, and bought more of it.

This paper sets the 2024 essays beside what happened. It does so in the manner of the house’s other public doctrine papers: it decodes the engineering, prices the structures, and states what would prove its reading wrong. Its conclusion can be given at once. The 2024 essays were right that a parametric trigger leaves losses uncovered, and wrong about what to do about it. A parametric trigger is not a measurement of damage. It is a contract on a measurement, and its worth lies in paying fast and without argument when the measurement says it must.

I. What the 2024 essays argued

The essays made three claims. The first was that central pressure alone was too narrow a variable: a storm can devastate without reaching the pressure the contract requires, and Beryl seemed to prove it. The second was that the geography was too rigid: boxes drawn in advance cannot follow a storm whose damage spreads across the region. The third was that the payout scale was too blunt: a linear schedule from 30 to 100 per cent of principal, keyed to one variable, could leave a country with nothing after a severe event that missed the threshold. The remedy proposed was a composite trigger — wind above 130 miles an hour, surge above fifteen feet, rainfall above ten inches in a day — with larger or adaptive boxes and payouts rising as more conditions were met. One of the essays went further and proposed real-time sensor data and automated settlement on a distributed ledger.

The essays also recorded the bond’s own terms as published at issue: an initial attachment probability of 2.34 per cent, an expected loss of 1.5 per cent a year, and a risk margin of 7 per cent, with the notes sold mostly to dedicated insurance-linked funds and asset managers in the United States, Europe and Bermuda.

II. What happened

Beryl, in July 2024, approached Jamaica as a Category 5 storm and tracked just south of the island. Its path did not carry it into any box at a pressure low enough to trigger the notes; the nearest qualifying box, at 960 millibars, was narrowly missed. Jamaica’s separate parametric policy with the Caribbean Catastrophe Risk Insurance Facility did respond. Jamaica’s Ministry of Finance put the point that the 2024 essays had missed in a single sentence:

“It is neither expected nor designed that all storms will trigger all instruments.”

Ministry of Finance and the Public Service, Jamaica, 2024

Melissa, in October 2025, was the storm the upper layer had been bought for. It crossed the island at a central pressure of 892 millibars, and the bond paid in full. The facility paid a further $91.9 million under Jamaica’s cyclone and excessive-rainfall policies. The cost of the damage, as estimated by the World Bank and the Inter-American Development Bank, was about $8.8 billion, or 41 per cent of Jamaica’s gross domestic product.

In May 2026 Jamaica placed IBRD CAR Jamaica 2026. The new bond keeps the parametric design the 2024 essays criticised: storm location and minimum central pressure from the National Hurricane Center’s best-track data, boxes laid out much as before, and a linear payout from 30 to 100 per cent, with any hurricane at 900 millibars or lower in any box able to trigger the full amount. What changed was quantity and price. The target of $150 million was oversubscribed and upsized to $200 million for four seasons. The attachment probability was set higher, at 3.86 per cent, and the expected loss at 2.48 per cent. The notes priced at a risk margin of 6.75 per cent, below the 7 per cent paid in 2024.

III. What a parametric trigger is for

An indemnity contract pays for the loss the insured actually suffered, after the loss has been measured. A parametric contract pays a pre-agreed amount when an independent measurement of the event crosses a pre-agreed threshold, whatever the loss turns out to be. The difference is not a defect of the second kind of contract. It is the reason it exists. A sovereign does not hold the granular exposure data an insurer holds, and much of what a hurricane destroys in a country like Jamaica — informal housing, uninsured livelihoods, public infrastructure — would take months to value. A parametric trigger replaces that valuation with a measurement that is published, independent and quick. The measurement is the contract.

The price of that speed is basis risk: the measurement and the damage can disagree. Beryl was a case of the measurement saying less than the damage. The engineering question is not how to make the measurement agree with the damage in every case, because no measurement will. It is how to build a set of instruments in which each layer is triggered by the events it was priced for, and in which the gaps between layers are known in advance. The upper layer of Jamaica’s structure was priced for an event expected about once in forty years. Beryl, by the estimate of one of the bond’s analysts, was closer to a once-in-five-years event at the island. It belonged to the lower layers, and the lower layers responded.

IV. Where the 2024 essays were right, and where they were not

2024 claim	What the record shows	Restated position
Central pressure alone can miss real damage	Correct: Beryl damaged Jamaica and missed the 960 mb box	Basis risk is real and must be managed across layers, not inside one trigger
Add wind, surge and rainfall to the same trigger	Each added variable needs its own measurement agent and model; rainfall was already covered by a separate parametric policy	Add perils as separate layers or tranches, not as conditions inside one contract
Enlarge or overlap the boxes	Larger boxes raise the attachment probability and the premium for the same cover	Box design is a pricing decision; it trades frequency of payment against cost
Pay more as more thresholds are crossed	A composite schedule would slow determination and widen disputes	Speed and certainty are what the sovereign buys; protect them
Real-time sensors and ledger settlement	The trigger that paid used best-track data and a calculation agent, and the cash arrived within weeks	Settlement was not the constraint
The design served investors, not the country	The design paid the country in full in its worst storm on record	The constraint is scale, not trigger design

The house withdraws the specific thresholds the 2024 essays proposed. They were offered without a model of how often each would be crossed, what it would cost to buy, or who would measure it. A threshold chosen because it describes a destructive storm is not yet a contract anyone can price.

V. The arithmetic of cover

The figures the bonds themselves publish allow the cost of protection to be read directly. In 2024 Jamaica paid a risk margin of 7 per cent a year on $150 million, about $10.5 million a year, for cover with an expected loss of 1.5 per cent. The investors were paid about 4.7 times the expected loss. In 2026 it pays 6.75 per cent on $200 million, about $13.5 million a year, for cover with an expected loss of 2.48 per cent: about 2.7 times the expected loss. After a full loss, the price of each unit of risk fell. The wider market moved the same way. By the end of the third quarter of 2026 the catastrophe bond market stood at about $65.5 billion outstanding, nine-month issuance had reached a record $18.9 billion, and the average spread in the third quarter stood at about 2.3 times expected loss, against 3.3 times across the first nine months of 2025.

The lesson in those numbers is that investors paid out exactly as the contract required and came back for more at a lower price. A trigger that pays when it should, and only when it should, is what makes that possible. The market does not punish a loss it priced. It punishes uncertainty about whether a contract means what it says.

The figure that matters most is the last one. Against damage estimated at $8.8 billion, the two parametric instruments together paid about $242 million — roughly 2.7 per cent of the cost. No redesign of the trigger would have changed that proportion materially. The protection gap after Melissa was a gap of scale, not of design.

VI. What a financial engineer would do

The house prices structures; it does not recommend which a government should adopt. With that reservation, the record points to five lines of engineering.

• Keep the upper layer parametric, simple and fast. Its value is that it paid the full amount within weeks of the worst storm in the island’s history.
• Widen the tower beneath it. More frequent events, like Beryl, need layers priced for their frequency — contingent credit, the regional facility, and domestic reserves — so that the cat bond is never asked to do a job it was not bought for.
• Add perils as separate tranches rather than as conditions inside one trigger. A rainfall or surge layer with its own measurement and its own price keeps each contract clean and each payout fast.
• Size cover to fiscal need. The binding constraint after Melissa was the ratio of cover to damage. Falling multiples and record capacity make larger cover cheaper per unit of risk than at any recent point.
• Pre-commit the use of the payout. Speed is only valuable if the money moves; reconstruction rules agreed before the season turn a fast payout into fast recovery.

VII. What would falsify this reading

This reading would be wrong if damaging storms that miss the upper trigger turn out to be common relative to those that meet it, so that the layer pays too rarely to justify its premium. It would be wrong if later sovereign deals priced after a loss harden rather than soften; the 2026 bond is one deal, and one deal is a reading, not a finding. And it would be wrong if fast payouts do not in fact reach reconstruction faster, because the case for a parametric trigger rests entirely on what speed buys.

VIII. The executed argument

Global Financial Engineering holds no insurance-linked securities and has no position in this market. This paper therefore reasons from public terms and public arithmetic rather than from the testimony of an executed trade, and it says so. The lesson it banks is one the house applies to its own machine as much as to a sovereign’s insurance: a rule should be judged by whether it does what it was written to do, not by whether it does everything one might wish. A trigger is a contract on a measurement. Jamaica’s paid when the measurement said it must.

Sources

• World Bank, press release, 7 November 2025: Hurricane Melissa triggers 100% payout of $150 million World Bank catastrophe bond for Jamaica.
• Artemis, deal directory and reports on IBRD CAR Jamaica 2024 and IBRD CAR Jamaica 2026 (May 2026), including trigger structure, attachment probability, expected loss and pricing.
• Artemis, reports of July 2024 on Hurricane Beryl and the Jamaica catastrophe bond, including the Ministry of Finance statement.
• Artemis, Catastrophe Bond and ILS Market Report, Q3 2026.
• Global Finance, on the damage estimate of the World Bank and the Inter-American Development Bank and the CCRIF payouts after Melissa.
• CLM Magazine, 29 October 2025, on Melissa’s landfall intensity.
• Insurance Business, November 2025, on market reaction to the payout.

GFE-PUB-DOC-20261017-A · Spirit Reigns · Substance Receives · Thought Impresses