A relatively benign start to 2026 has given the global insurance industry a measure of stability. But the changing nature of catastrophe losses, rather than their scale, is prompting a rethink of how risk is modeled.
Insured losses from natural catastrophes reached at least $20 billion in the first quarter, according to Gallagher Re, well below both recent and long-term averages. Economic losses were estimated at $58 billion, underscoring a persistent protection gap.
For now, the absence of a major loss event has left insurers and reinsurers well positioned ahead of what is typically a more volatile mid-year period, said Steve Bowen (pictured), chief science officer of Gallagher Re. “We did not have a $10 billion, $20 billion or larger event that would put the market on edge going into peak season.”
For Bowen, catastrophe (cat) models are being pushed beyond their traditional limits as non-hazard factors increasingly determine the cost of disasters.
The clearest illustration of this shift is the sharp rise in losses from US severe convective storms such as hail, tornadoes and straight-line winds. Since 2008, insured losses from such storms have grown markedly, with several years exceeding $30 billion and, more recently, $50 billion. While climate variability has played a role, Gallagher Re’s analysis suggested that the bulk of the increase is not meteorological.
Instead, between 80%-90% of the long-term rise in losses can be attributed to economic and societal factors: higher construction costs, labour shortages, supply chain disruptions and a more litigious claims environment.
“The assumption has often been that weather is the primary driver,” Bowen said. “But when you look closely, it is the cost of rebuilding and the way claims are handled that are having the biggest impact.”
Bowen cited the price of asphalt roofing, widely used in the US, as a case in point. Derived from oil, it became significantly more expensive during the commodity price surge of the late 2000s. While crude prices later eased, construction materials did not fully follow, embedding higher costs into the system.
The pandemic amplified these pressures, he said, with supply chain shocks and inflation pushing up both material and labour expenses. At the same time, practices such as assignment of benefits and increased litigation have driven up loss adjustment costs.
This confluence of factors is challenging the foundations of cat modeling, which has traditionally focused on the probability and severity of physical hazards.
“It is no longer just about whether a building is damaged. There are multiple layers that determine the final loss, such as economic conditions, regulatory frameworks, and even geopolitical issues that affect reconstruction costs,” Bowen said.
At the same time, underlying exposure continues to rise. Population growth in high-risk regions has increased the number of assets at risk, particularly in parts of the US prone to convective storms. Over the past quarter-century, millions of new homes have been built in such areas, amplifying potential losses.
While many of these variables are well understood individually, they have not always been systematically incorporated into models used for underwriting and capital allocation. The result is a growing gap between modeled losses and realized outcomes.
Bowen believes the industry is now moving, albeit gradually, towards more integrated approaches that attempt to capture these broader dynamics.
“These factors have been understood at a macro level, but not always integrated into traditional catastrophe modeling,” Bowen said.”There is a growing level of sophistication, and we hope this kind of analysis helps show how to think about risk more holistically.”
Technological change is also adding to the complexity. Residential properties increasingly include solar panels, battery storage systems and other equipment that raise repair costs.
On the commercial side, the expansion of data centers, fuelled by demand for artificial intelligence, has created concentrations of high-value assets that are vulnerable to weather-related disruption, said Bowen.
“Many SCS-prone states do not have stringent (building) codes, which is concerning. Companies building these facilities have to decide whether to meet minimum standards or build beyond them to withstand higher wind speeds or hail risks,” he said.
“Given the size and value of these facilities, even partial damage can lead to very large claims. It’s similar to what we’ve seen with major commercial losses in manufacturing or pharmaceuticals. As more of these facilities are built, the potential risk increases.!
According to Gallagher Re’s analysis, the insurance industry’s capital position has strengthened in recent years, leaving it more resilient to shocks than in previous cycles. Bowen estimated that insured losses would need to exceed expectations by more than $115 billion before pricing dynamics in the reinsurance market are materially affected.
However, the next generation of cat models will need to incorporate a wider range of inputs, from macroeconomic conditions to demographic trends and political developments, in order to better capture risk.
“The industry has to broaden its lens,” Bowen said. “Understanding the science of the hazard remains critical, but it is only part of the picture. We’re not saying climate change or weather volatility doesn’t play a role, because it absolutely does, but it challenges the idea that it’s the sole driver. These non-hazard factors are driving the bulk of increased loss costs."
