Catastrophe insurance: the market El Niño is accelerating

91% of Brazil's climate losses have no insurance coverage of any kind. The figure comes from CNseg/EY, presented at COP30 in 2025. It is not a failure of market awareness. It is a failure of instrument.

The El Niño confirmed for 2026/2027 is doing what no report alone could: accelerating the conversation about catastrophe insurance in Brazil. That leaves a simple question. Why do we need a climate phenomenon knocking at the door to solve a problem that had already been sitting in the numbers for years?

What El Niño is actually accelerating

In June 2026, the insurance sector registered a signal of changing posture. Gente Seguradora reported that, after El Niño's confirmation, insurers, industry associations and public agencies advanced their mobilization for catastrophe insurance and climate protection measures in Brazil. The topic moved out of the seminar room and onto the boardroom table.

This lines up with what the press had already reported in January of the same year: the intensification of extreme weather events, from urban flooding to prolonged rural droughts, is forcing Brazilian insurers to rethink protection models, risk management and pricing. Revising a model is not compliance routine. It is an acknowledgment that the old model cannot handle the new risk.

The new risk has global scale, but the figure deserves precision, because this is where an important point lives for anyone working with traceable data. Munich Re, one of the world's largest reinsurers, recorded US$ 112 billion in global losses from natural disasters in the first half of 2026, of which around US$ 44 billion were insured. That is, in fact, a milder half than average: below the inflation-adjusted 10-year average (US$ 113 billion in total losses, US$ 50 billion insured) and the 5-year average (US$ 136 billion and US$ 66 billion).

The point of attention is not the first half's volume. It is what Munich Re itself is signaling for the second half. Tobias Grimm, the reinsurer's chief climate scientist, described the combination as dangerous: global warming continues and the world is heading toward a Super El Niño, which should push temperatures even higher, with effects concentrated in the second half of the year. Thomas Blunck, a Munich Re board member, called the first half "a welcome respite" after years of elevated losses, but reinforced that climate change and growing exposure keep raising the risk of larger losses ahead.

When reinsurance talks about future risk instead of celebrating the present result, the whole market feels the signal. It is reinsurance that carries the tail risk local insurers cannot hold on their own.

El Niño did not create Brazil's catastrophe insurance problem. It just pulled it out of the closet faster.

91% uncovered: the number that opens the debate

It is worth unpacking the initial figure. CNseg/EY calculated R$ 184 billion in climate losses in Brazil between 2022 and 2024, across 67 significant events, an average of R$ 60 billion per year. Looking at the more recent window, from 2022 through June 2025, the total rises to about R$ 215 billion across 77 events. Of that accumulated total, 91% had no insurance coverage whatsoever.

A separate study, covering the same period, arrives at a similar reading through a different path: Brazil loses about R$ 110 billion a year to climate disasters, and only 10% of the losses are insured. The methodologies are not identical, but the order of magnitude and the direction of the problem are confirmed by independent sources: most of Brazil's climate loss simply is not insured.

The year 2024 illustrates what that number means in practice. The May event in Rio Grande do Sul caused R$ 89 billion in direct losses, affected 2.4 million people, cost 182 lives and closed Salgado Filho Airport for months. Even as the year of reference for the tragedy, insurance coverage for the state stood at 8%, about R$ 7.3 billion paid out. Compared to developed countries, where between 20% and 55% of climate losses are covered, Brazil runs at 9%.

The distribution by insurance line in 2024 shows where existing protection is concentrated: property insurance accounts for 58% of climate claims, auto for 19%, rural and agribusiness for 15%, homeowners for 6%, and the remaining lines together for 2%. In the North and Northeast regions, climate coverage sits below 2%. In homes, 15%. In planted agribusiness area, less than 5%.

It is not a lack of product. Brazilian insurers have a policy for practically every type of asset, from a port to a power plant, from a grain silo to a transmission line. The bottleneck is elsewhere: without reliable physical data on the specific risk of each asset, the product cannot be priced with confidence. A product that cannot be priced right either never leaves the drawing board, or ends up too expensive for the client, or too cheap for the insurer to sustain in the long run.

The bottleneck is not the event, it is the risk model

Here lies the point that usually goes unnoticed in conversations about catastrophe insurance. The problem is not the climate event itself. Floods happen, droughts happen, heat waves happen. That has always been true. The problem is that a large part of the insurance industry still prices climate risk with regional-resolution data, something in the range of 25 to 100 km grid cells, when the asset that needs coverage sits at a specific point on the map: a port, a substation, a stretch of railway, a farm plot.

Climate risk is not uniform within a 25 km area. A port can be exposed to a wind and tide dynamic completely different from the neighboring city, 20 km away. A power plant can sit in a microclimate of intense rainfall that the regional average simply dilutes. A farm can face a drought regime that the municipality next door does not feel in the same way. When the pricing model sees the region but not the asset, it systematically errs both ways: it underprices what is genuinely at risk and overprices what is not, or it simply refuses to insure what it cannot measure.

The frequency and severity of extreme events have also been changing in Brazil. The Atlas Digital de Desastres no Brasil, maintained by the National Secretariat for Civil Protection and Defense (Sedec/MIDR) in cooperation with UFSC, gathers official disaster records across the country between 1991 and 2025, and shows a clear upward trajectory in the frequency of recorded events over recent decades. The event is more frequent. The pricing model, historically, has not kept pace with the same rate of refinement.

Two sides of the same problem: underwriting and disclosure

The granular physical data bottleneck does not show up only in insurance pricing. It also shows up on the other side of the balance sheet, in the climate risk disclosure of insured companies.

Since May 2026, CVM Resolution 244 made climate risk reporting voluntary for Brazilian publicly listed companies, reversing the mandatory requirement that had been in effect. The current regime operates under a "comply or explain" model, with no new deadline set for an eventual return to mandatory reporting. CVM Resolution 218, from 2024, remains the technical reference standard, not a requirement.

This does not eliminate the pressure. Of the roughly 700 publicly listed companies on B3, only 8 had formalized voluntary adoption of the standard by the end of 2025. For the financial sector, the bar is different: banks and conglomerates remain bound by the Central Bank's track (CMN 5.185 / BCB 435), regardless of CVM's decision.

The connection point with catastrophe insurance is direct. Both the underwriter pricing a policy and the CFO signing off on a climate risk report need to answer the same question: what is the real exposure of this specific asset, not of the region where it sits? Without fine-resolution, per-asset data cross-referenced with future climate scenarios, both answers stay generic. And a generic answer sustains neither defensible insurance pricing nor defensible disclosure under audit.

What changes when the data matches the size of the asset

The way out is not more alarm, it is more resolution. A physical risk model that works at the same grain as the insured asset, rather than the average of an entire region, changes what an insurer can do on three fronts.

In underwriting, it allows pricing with today's climate and with future scenarios calibrated per asset, instead of the historical average of a broad grid. This reduces both the underpricing of assets under real risk and the refusal to insure what was simply never measured properly.

In portfolio management, it allows anticipation by event, not just by season. Knowing that a cold front or an extratropical cyclone will hit a specific region of the portfolio gives time to mobilize response, communicate with highly exposed policyholders and adjust aggregate exposure before the event, not after the claim has already hit the books.

In claims and provisioning, it allows building an evidence trail: which climate variable was observed, at what time, at what location, at what intensity. That trail supports cause analysis, subrogation against third parties, force majeure evidence and defense in legal disputes, with traceable method instead of a standalone technical opinion.

The common denominator across all three fronts is the same: fine-resolution physical data, calibrated for Brazil and Latin America, not generic global data forced to fit. That is the bottleneck separating a catastrophe insurance market that reacts to the event from one that prices it with method before it happens.

El Niño is not the problem. It is the deadline

Brazil's catastrophe insurance market was not waiting for El Niño to exist. It was waiting for the data it lacked to work properly. The 2026/2027 climate phenomenon only made visible, with greater urgency, a problem i4sea had already seen in the numbers before: Brazilian climate risk is real, measurable and increasingly frequent, but it keeps being priced with the wrong resolution.

For insurers, reinsurers and companies with assets exposed to physical climate risk, the question worth asking now is not whether the next event will happen. It is whether the data used to measure it sees the asset, or only the region.

Request a free climate exposure assessment

Register an asset, receive a physical risk snapshot within 48 hours using proprietary data at 1 to 3 km resolution, and see the difference between regional estimate and per-asset evidence.