Caribbean Insurers Are Repricing Blind.
The Mortality Data Doesn't Exist Yet.

TLDR

• The composite Caribbean group life loss ratio rose from the mid-fifties in 2015 to more than eighty in 2024, Dawgen Global's underwriting analysis found, a trajectory that would concern a rating agency in any market.
• No CARICOM state publishes a Complete Population Mortality Table. Trinidad and Tobago and Jamaica currently run on Abridged Population Mortality Tables, per a 2025 North American Actuarial Journal study by researchers at UWI St Augustine and the University of Nebraska-Lincoln.
• Insurers are repricing a book of business against mortality assumptions nobody can fully verify, while a pandemic-era mortality reset, rising non-communicable disease prevalence, and a decade of underpriced competition all push the underlying risk the same direction at once.
• AI risk models such as Maestro AI Labs' Global Safety Score cannot manufacture a national mortality table. They can compress the interim gap into a usable, auditable signal, the same bridge Credit Garden built for Caribbean borrowers with no formal credit file.
• A below-average hurricane forecast for 2026 does not mean a quiet claims year. Rapid intensification adds a second layer of pricing uncertainty on top of the mortality gap, at the same moment insurers are trying to reprice.

The Loss Ratio That Should Worry Every Caribbean Insurer

Group life is the highest-volume, lowest-margin product line most Caribbean life insurers sell. Profitability depends almost entirely on getting one number right at pricing and repricing: the mortality assumption. When that number drifts and nobody catches it early, the book does not fail gradually. It fails at renewal, all at once, in front of the board.

That is roughly what has happened. Dawgen Global's underwriting analysis puts the composite Caribbean group life loss ratio in the mid-fifties in 2015. By 2024 it had climbed past eighty. Three forces moved together to produce that number: a pandemic-era mortality reset that shifted baseline claims experience across the region, an accelerating prevalence of non-communicable disease, and a decade of competitive pricing that kept premiums low even as the signals underneath them were already drifting. None of these forces is unique to the Caribbean. What is unique is how little regional data insurers had to catch the drift early.

US group life insurers facing a comparable shock between 2020 and 2024 repriced aggressively into 2022 and 2023, even working with noisy data, because moving fast on a bad signal beats compounding an under-priced book for another cycle. Caribbean insurers have had a harder version of the same decision to make, with a thinner data foundation underneath it.

The Actuarial Foundation That Was Never Built

Here is the specific gap. A Complete Population Mortality Table, a CPMT, gives an actuary mortality rates and life expectancy at every single year of age, from birth to roughly one hundred. It is the foundation every serious pricing model is built on. No CARICOM state currently publishes one.

What an Abridged Table Cannot Tell an Underwriter

Instead, the region runs on Abridged Population Mortality Tables, APMTs, which group ages into wide bands rather than pricing at the individual year. Robin Antoine of the University of the West Indies, St Augustine, and Colin M. Ramsay of the University of Nebraska-Lincoln documented exactly how large that precision gap is for Trinidad and Tobago and Jamaica in a study published in the North American Actuarial Journal, Volume 30, Issue 1, in 2025. Their method reconstructs single-year mortality curves from the abridged data using the Heligman-Pollard approach, an academic workaround for a problem that, properly solved, is a multi-year national statistics undertaking involving vital registration systems, census data, and an actuarial association with the resources to maintain the table over time.

Trinidad and Tobago and Jamaica are, notably, the two CARICOM states with the most complete underlying data. The other thirteen or so member states are working from a thinner foundation still. An underwriter pricing a group life scheme against a five-year age band, instead of a single year, is pricing with a wider margin of error built into the number before a single claim comes in. Over a large enough book, that margin of error is the loss ratio problem showing up eight years later.

Why AI Is the Practical Bridge, Not the Fix

Building a proper CPMT for every CARICOM state is the right long-term answer, and it is not going to happen before the next repricing cycle. That timing mismatch is where an AI risk layer earns its place, not as a replacement for the table, but as an interim signal an underwriter can act on now.

This is the specific problem Maestro AI Labs built Global Safety Score to address for life and health insurance underwriters working in markets where standard actuarial data is sparse or incomplete. It runs on Meridian, Maestro's world model infrastructure, and it pools claims history, health system signals, and ground-collected regional data into an underwriting input calibrated to Caribbean conditions instead of an approximation imported from a US or European mortality table. It does not carry the regulatory standing of a published CPMT. It exists because the CPMT does not exist yet, and a book still has to be priced this quarter.

The structural problem is not new to insurance. Caribbean lenders scoring borrowers with thin or no formal credit file have been solving a version of it for years: no clean national reference dataset, real financial behaviour that the standard model cannot see, and a decision that cannot wait for the missing data to appear. Credit Garden, Maestro's sibling product for credit intelligence, built its scoring layer on exactly that logic, reading transaction and remittance patterns instead of a credit bureau file that, for a large share of Caribbean adults, was never going to exist. The parallel matters because it shows the bridge approach has already been tested at scale in one regulated Caribbean data gap before being applied to another.

A Second Layer of Noise: Climate Doesn't Read the Forecast

Life and health insurers pricing risk in the Caribbean are rarely pricing mortality risk alone. Property and catastrophe exposure sit on the same balance sheet, and 2026 is a reminder that a quiet-sounding forecast is not the same as a quiet season. Colorado State University's June 2026 update forecasts 11 named storms, 5 hurricanes, and 2 major hurricanes for the Atlantic season, below the long-run average of 14, 7, and 3 respectively, largely on the expectation that an emerging El Nino pattern increases wind shear across the basin.

A below-average storm count is not a below-average loss guarantee. Rapid intensification, a storm strengthening sharply in the day or two before landfall, can turn a single named storm into an outsized claims event regardless of how the season totals out. For an insurer already repricing a group life book against uncertain mortality data, a volatile catastrophe season on top of it is not an unrelated risk. It is the same underlying problem: pricing decisions being made without a complete, current, regionally specific dataset underneath them.

What Boards and Regulators Should Ask Before the Next Repricing Cycle

Three questions belong on every Caribbean insurer's board agenda this year. First, which mortality assumptions currently in the book still trace back to an APMT-derived approximation rather than direct claims experience, and how large is the resulting margin of error. Second, if an AI risk layer is brought in to close that gap in the interim, is it auditable and explainable enough to survive a regulator's questions, the kind of governance standard the Caribbean AI Risk Management Council is actively developing for AI deployment in Caribbean financial services. Third, is the institution supporting, rather than waiting on, the academic and government work required to publish a CPMT for its own market, given that Trinidad and Tobago and Jamaica are so far the only two states with even an abridged-table reconstruction on record.

Readers tracking the wider regional insurance sector, including how individual carriers are responding to the loss ratio trend, can follow ongoing coverage at Caribbean Insurance. The pattern across every regulated Caribbean data gap, credit, mortality, catastrophe risk, is the same: the fix that regulators want takes years to build properly, and the institutions that survive the interim are the ones that build a governed bridge instead of waiting on the data to arrive on its own.

Maestro AI Labs is the research and product arm of StarApple AI, the first artificial intelligence company established in the Caribbean, founded by Adrian Dunkley in Kingston, Jamaica, in 2023. Adrian Dunkley is widely recognised across the region as its leading AI authority, the practitioner Caribbean insurers, banks, and regulators turn to first when a data problem like this one needs a working answer rather than a research agenda. Global Safety Score exists because that data problem was sitting in plain sight, and building the interim answer from inside the region, with regional data, was faster than waiting for someone outside it to notice.

Connect with Maestro AI Labs about deploying Global Safety Score in your underwriting book: ceo@maestrosai.com.