Monsoon Misery Floods In

A study published in Nature in November shows that the number of deaths reported as caused by rain and floods is usually considerably fewer than the actual. In countries from Bangladesh to the USA, many more people are killed by flood events than is recorded in official data.

But already the numbers are high, even for hyper-localized events: floods in Valencia a year ago killed a reported 229 and cost €10 billion. This July, 135 died, including many children, when the Guadalupe River flooded in Texas.

It is late in the monsoon season in Southeast Asia. During this waning rainy phase of the region’s regularly shifting annual wind pattern, three embedded tropical cyclones formed. The rains poured down from warmer-than-average skies.

This year’s monsoon has been unusually intense and prolonged, which has led to greater overlap with an extended cyclone season. Monsoon-season rainfall is always heavy, but last week it was intensified by the unusual cluster of storms. The three cyclones formed on successive days over the Bay of Bengal (Cyclone Ditwah), the Malacca Strait (Cyclone Senyar) and the South China Sea (Cyclone Koto).

The storms’ combined impact has been catastrophic. They helped funnel vast amounts of water into huge areas of Southeast Asia, some of which were quickly inundated. Sri Lanka, for example, suffered more than 300mm of rainfall an hour over a 48-hour period, prompting the President to describe “the entire country” as “struck by disaster.” Thailand’s Hat Yai suffered a 300-year downpour, driving floodwaters as high as eight feet through the city’s streets.

The situation was made worse by La Niña conditions, which currently are reinforced by a positive phase of the Indian Ocean Dipole (IOD). The dipole is an irregular ocean circulation driven by differences in sea-surface temperatures. When in its positive phase, the western Indian Ocean becomes warmer, and the east cooler. That causes increased rainfall in the west, and drought in the east. However, the positive IOD pattern tends simply to shift rain towards Southeast Asia, where the floods occurred.

Together these factors created an exceptionally wet background state. When three cyclones hit at once, disaster was inevitable.

The Bay of Bengal and the western North Pacific are in peak cyclone season, but the clustering stands out. So too does the formation of Cyclone Senyar in the Malacca Strait, near the equator, where cyclones hardly happen.

The intensity reached by the three storms is also unusual. Each developed under conditions that would not typically support such powerful cyclones, especially near the equator. Climate change may be exacerbating the phenomenon by altering the boundaries of where storms can form and intensify.

The ocean was exceptionally warm in the run-up to the cluster event. This year’s temperatures follow the long-term warming trend fueling higher sea-surface temperatures, which provide the heat and moisture that feeds storms. In addition, warmer air holds more moisture. Together these factors can lead to heavier and more extreme rainfall, which in the events of the past two weeks has brought on severe flooding.

The events align with recent findings from the National University of Singapore, a Willis Research Network partner. The University’s high-resolution simulations of 2024’s Typhoon Yagi showed that even one to three degrees centigrade of sea-surface warming can significantly intensify storm rainfall and winds. Their work demonstrates how small temperature changes enhance convective buoyancy and energy fluxes. It underscores how even modest warming can reshape storm behavior and risk profiles.

No individual cyclone can be directly attributed to climate change. However, last week’s storms illustrate how warming oceans and shifting climate patterns are making extreme events like Ditwah and Senyar more intense, clustered and difficult to predict.

Reinsurance markets hesitate over flood risk because it has been notoriously difficult to model. In markets where underlying policies typically cover the peril, flood events could and regularly did cause hundreds of unanticipated, and therefore unpriced, claims that cut a swathe through reinsured portfolios. Until recently, the major model vendors were largely silent on flood, while reinsurers’ first-generation in-house models required further calibration.

Meanwhile, outdated flood maps continue to pass judgement on the vulnerability of huge areas of territory. For example, reThought Flood’s study of April’s Ohio Valley flood event found that 6,932 flooded properties, or 32% of the total hit by the event, lie in FEMA’s low-risk “X Zones.” But these maps may still guide decision-making about a peril where centimeters matter.

Technology is beginning to help. The ability to collect, parse and analyze massive datasets, combined with financing mobilized by the insurTech investment boom, has prompted the emergence of several specialist flood modelers and underwriters, particularly MGAs. These advances have made flood a much more palatable risk for carriers across the spectrum.

Still, flood remains one of the perils which keeps reinsurers awake at night.

In the US, the formation of the National Flood Insurance Program, or NFIP, in 1968 allowed flood to be easily and comfortably excluded from almost all homeowners and commercial insurance policies. That took it out of primary carriers’ outwards treaties.

Under the US model, only those most at risk bought flood insurance, typically guided by the blunt instrument of the Federal Emergency Management Agency’s nationwide flood maps. Lenders still commonly use them to determine the necessity of standalone flood coverage for mortgaged properties, which is typically provided by the NFIP, which is tens of billions of dollars in debt.

A private commercial flood insurance market has evolved in the US, driven by model improvements. It is beginning to thrive. Recent accelerants include the record-long federal government shutdown, which prevented the NFIP from issuing new policies. Another is the favorable pricing which may now be offered to less vulnerable properties despite their map coordinates, after new technologies have reassessed their risk profiles. Portfolios of such cherry-picked flood risk are easily reinsured.

In Europe, flood is much more likely to be included in typical homeowner and commercial policies, although it is carefully underwritten, often sub-limited, and regularly carved out based on loss record and/or proximity to bodies of water. Still, a pair of flood clusters in 2024, one in southern Germany and one further south, cost insurers a total of €3.7 billion, according to PERILS.

The German portion of that loss illustrates flood-model variance. PERILS ultimately placed the insured loss in Germany at €1.6 billion. Verisk, successor to AIR Worldwide, had pegged the loss in the range of $2.6 billion to $3.9 billion, however, and Moody’s RMS calculated an upper limit of $3.2 billion. No wonder reinsurers are nervous about flood.

Even more costly were the floods arising from storm Bernd in 2021, which soaked areas of Germany, Belgium and the Netherlands to cause the largest-ever insured European flood loss, in excess of $13 billion. Much of the loss fell to reinsurers and their retro programs. It was one of the events that caused the market retrenchment of European excess of loss reinsurance after a prolonged soft market.

Occurrence and aggregate treaties typically include coverage for European flood events, as do many property retro covers, although some markets will say an outright “no” to flood. In general, reinsurers are now becoming slightly more generous with capacity, as the market softens. Excess of loss capacity is now available to attach slightly lower down programs. However, it rarely reaches the working layers insurers may desire, and flood sub-limits are often requested.

A growing appetite for parametric flood reinsurance is available to plug this gap. Treaty protection can be structured to respond to a single large flood event, or to a modeled aggregate loss arising from smaller events which threaten to erode insurers’ earnings. It is also proving popular among captive insurance company managers who often find that, for example, goods held at riverside distribution points are very difficult to insure against flood. Meanwhile, facultative markets for well-modeled risks are thriving as competition begins to rise in this arena, and markets begin to soften.

All of this follows a period when cedants leaned heavily on reinsurers to cover flood losses. In many cases, particularly following multiple reinsured flood loss events in 2024, flood risk had not been priced adequately into aggregate reinsurance programs that nonetheless responded to multi-billion-euro flood events.

Several countries have state-backed flood insurance programs for properties which the private market will not insure under conventional policies. These national flood insurers of last resort are important buyers of commercial-market reinsurance capacity, and therefore provide a market bellwether.

The US NFIP reported a 2025 cession of $757.835 million of risk to 27 global reinsurers (up from 18 the previous year). The coinsurance tower covers a single occurrence and attaches at $7 billion, providing coverage of 12.0334% of losses up to $9 billion, and 25.8584% of losses between $9 billion and $11 billion (for reference, NFIP’s ultimate loss cost for Hurricane Helene was estimated by the agency to fall between $6.4 billion and $7.4 billion). The program cost FEMA $139.9 million in 2025. In addition, it has in-force reinsurance worth nearly $1.25 billion through three FloodSmart Re Ltd. catastrophe bonds.

In the UK, state-backed Flood Re (no relation) assumes 80% of flood risk sold under coverage attached to conventional homeowners’ policies against a state guarantee and a conventional commercial reinsurance program. At its April 2025 renewal, Flood Re nearly tripled its retention to a reported £347 million, and declared that its reinsurance program cost had increased by more than £100 million per year.

The reinsurer’s CEO warned that the global reinsurance market’s appetite for UK flood risk is sated, although well-modeled risk can always be ceded at an adequate price. The tools at our disposal can help to narrow down divergent views of risk and build the confidence markets need to sign the line.