Two sides of the same coin: turning science into better risk decisions

Record reinsurance capital is giving European insurers more choice over how they buy protection, but analytics will be critical in matching that capacity to long-term business objectives, says Willis Re’s Tina Thomson.

Key points:
Science sharpens view of emerging risk
Analytics turns risk insight into decision
Brokers can link both with placement

By Dr. Tina Thomson, head of data and analytics, Willis Re

European insurers approach the 2027 renewals from a position of strength. Reinsurance capital, including alternative capital and catastrophe bonds, stands at record levels, and rates have continued to soften through 2026.

Yet even in a well-capitalised market, shock losses are being watched most closely, as a single large event, or clusters of events, can move a year’s result. Insurers that understand and can evidence their exposure are best placed to secure the protection they want, on their terms.

Evidencing it takes two things. The first is science: models combining the latest peril research with high-quality, current exposure data. The second is analytics: an actuarial, technology enabled approach connecting that view of risk to decisions on capital, reinsurance, portfolio optimisation and pricing. Together they give cedants and reinsurers a view of risk both sides can trust.

We have designed our analytics, data and technology on a modern platform, integrated with broking from the start. Catastrophe modelling, actuarial analysis, pricing, structuring and market data share one platform, giving analysts and brokers a single evidenced view of each client's portfolio. With AI-assisted data cleansing and a full audit trail, reinsurers can see exactly what has changed in a submission and why, which is the foundation of that trust.

Understanding the science

Wildfire, for example, is a peril where the science is moving quickly. Recent events have highlighted rapid fire intensification, where fires escalate from manageable incidents to major loss events within hours.

Dr. Matthew Jones, of the University of East Anglia’s (UEA) Tyndall Centre for Climate Change Research, co-lead author of the annual State of Wildfires report, works with us through the Willis Research Network (WRN). “The world is experiencing really drastic, fundamental changes in the global geography of wildfire,” he told a recent Willis Re webinar. “Fires are increasingly encroaching upon the built environment.” He points to “hydroclimatic whiplash”: intensifying swings between drought and heavy rainfall. Vegetation grows vigorously in the wet spells, then dries out, turning landscapes into fuel.

The current work with UEA is evaluating vapour pressure deficit (how strongly the atmosphere draws moisture out of vegetation) as a wildfire hazard metric alongside traditional fire risk indices. A physically grounded measure of dryness helps distinguish between locations that traditional indices treat alike, sharpening risk selection, pricing and accumulation management, and informing how much wildfire risk to retain or transfer.

“Understanding what drives a loss, not the headline, allows risk to be priced on its merits.” 

Testing the models

In a current proof-of-concept, we are testing a new-generation, physics-based wildfire model by re-running historical events, including the January 2025 California wildfires, under alternative conditions: ignition point, wind strength, fuel dryness, timing and development at the wildland-urban interface.

Moving the ignition point by two kilometres almost doubled the modelled loss. Limiting development at the wildland-urban interface to its 1975 extent cut it by more than a third. Re-run a month later as the January 2025 fire caused no loss at all: the vegetation had already burned. Counterfactual analysis turns one loss into a distribution of plausible outcomes. This is needed to test whether a model is fit for purpose and to set retentions and limits with confidence.

The same discipline applies to Europe’s convective storm risk. WRN has partnered with the Karlsruhe Institute of Technology (KIT) for more than a decade, producing the first Europe-wide stochastic hail model, with a higher-resolution update due at the end of October. KIT’s analysis of June’s heatwave, co-authored by our Willis Research Fellow, found that despite a highly unstable atmosphere, most thunderstorms stayed small and short-lived. Extreme heat does not automatically mean extreme hail. Understanding what drives a loss, not the headline, allows risk to be priced on its merits.

From science to decisions

The second side of the coin brings portfolio, reinsurance and capital decisions into one framework. Building a client’s objectives into an economic capital model lets us test how different reinsurance structures and alternative capital affect earnings volatility, solvency and growth. Risk scoring and loss-trend analysis feed back into pricing, so capital management becomes continuous rather than an annual event. 

With more options available in the market, analytics lets each client test whether a lower retention pays for itself, where frequency protection adds most value, and which mix of traditional and alternative capital is most efficient. It gives insurers insight to manage and price their own risk and reinsurers the confidence to support them.

The role of the broker is also evolving. Our ambition is to give every client actionable insight and a distinctly better experience of working with a reinsurance broker. For European insurers building long-term security in a changing risk landscape, both sides of the coin matter.

For more news from Baden-Baden Today, click here.

Did you get value from this story? Sign up to our free daily newsletters and get stories like this sent straight to your inbox.