The 1 January renewals is a key moment in the reinsurance calendar, with around half of global reinsurance business renewing at this time. But as the market has softened, with property-catastrophe rates falling by double digits at the 2026 renewals, the focus is not simply on price.

Risk has not become less complex. For insurers, demonstrating the quality of a portfolio and the evidence behind its risk view can be just as important as securing competitive terms.

Let’s take flood for example. A modeled result can provide valuable insight into portfolio exposure, but confidence in that result depends on the data, science and assumptions behind it.

With renewals approaching, here are some questions worth considering when assessing flood risk – from the quality of the underlying exposure data to how the model represents, validates and explains the risk.

1. How accurate is the underlying exposure data?

High resolution is valuable, but it is only one part of the picture. The quality of underlying data and science is ultimately what determines how useful the model is for risk selection and pricing.

The usefulness of any result starts with the exposure being modeled. Property values, locations and other exposure characteristics can change throughout the year. A good starting point is to consider whether the portfolio data remains current and whether properties have been accurately geocoded.

This matters particularly for flood, where relatively small differences in location can change the physical characteristics of a risk.

2. Does the model capture the flood perils that matter?

Flood is not a single peril. Fluvial, pluvial and coastal flooding have different physical drivers and can create very different patterns of exposure. Models that blur these distinctions create critical pricing and risk selection blind spots.

• Pluvial (flash): High-intensity rainfall can cause surface flooding far from traditional floodplains. Without granular surface runoff modeling, underwriters face frequent, unpriced low-severity claims in supposedly “safe” areas.
• Fluvial (riverine): Major river basins are standard, but models that truncate minor waterways leave portfolios exposed to unmapped flash flood risks.
• Coastal (storm surge): Marine surges and tides drive extreme tail-risk. Modeling coastal surge separately is essential for evaluating coastal exposure and avoiding accumulation surprises when surge blocks inland drainage.

Explicit sub-peril separation — backed by comprehensive river network coverage down to small streams — allows (re)insurance teams to refine risk selection, protect technical margins and allocate capital with confidence.

3. How are flood defenses represented in the model?

Evaluating flood defenses is a critical, often overlooked component of technical underwriting. Models that fail to accurately account for defenses introduce structural mispricing — either by overestimating hazard in defended locations or underestimating tail-risk when structures fail.

Defense representation impacts three key decisions:

• Pricing accuracy: A 1-in-100-year defense reduces frequent, minor losses, but residual risk remains. Access to both defended and undefended views allows underwriters to price residual risk accurately and set appropriate deductibles.
• Refining risk appetite: Blanket exclusions in defended zones mean missing profitable business, while blind acceptance exposes books to unpriced risk. Clear visibility into defense standards enables smarter, location-specific risk selection.
• Managing portfolio accumulation: Defenses alter loss patterns. When water levels exceed defense standards, inundation is rapid and severe. Capturing defended capacity alongside undefended hazard prevents major accumulation surprises.

4. Can you see what sits behind the model output?

A modeled loss or risk score provides an answer. A transparent methodology provides the context behind it.

Understanding the data sources, modeling approach and key assumptions can give greater confidence in how an output has been produced. It can also make it easier to explain and challenge when the result does not align with your wider view of a risk.

This is the principle behind Fathom’s glass-box approach. Its methodologies are developed and documented by Fathom’s science team, with research published through peer-reviewed academic literature. Methodologies and supporting information are made available, giving users greater visibility into how flood risk is modeled.

5. How has the model been validated?

Model sophistication is only one piece of the puzzle. Confidence also comes from understanding how a model has been tested. Validation against observed flood events, independent datasets and alternative models can provide useful evidence of how the model performs and where its limitations may lie. Look for open documentation provided by the vendor (or ask your cat team about how it has been validated).

Benchmarking can be particularly valuable when assessing large or complex exposures, or when modeled results have material impact on portfolio strategy or renewal terms.

No model will reproduce every flood event perfectly. Validation helps provide context for interpreting the result and understanding where further analysis may be useful.

6. Where does uncertainty sit?

Uncertainty is inherent in flood modeling and comes, to some degree or other, from every component spanning exposure and terrain data through to hydrological assumptions and the representation of flood defenses.

The useful question is not whether uncertainty exists, but where it exists. Which data inputs and geographical areas are more uncertain than others? Understanding this will allow you to appropriately factor it into a decision.

Greater transparency around the sources of uncertainty can help distinguish between a result that provides strong evidence of risk and one where additional investigation may be warranted. For portfolio management, pricing and capital allocation, that context can be as important as the modeled result itself.

Learn more about Fathom’s approach to uncertainty >

From model insight to underwriting decisions

The considerations above have a practical role in underwriting and reinsurance, where flood intelligence can help inform risk selection, pricing and portfolio management.

Swiss Re acquired Fathom in 2023 to strengthen its flood risk capabilities. Fathom’s flood hazard data is now integrated into Swiss Re solutions including CatNet, Portfolio Insights and Risk Data & Services, as well as being used within Swiss Re’s internal flood risk modeling.

Fathom data within CatNet gives users additional high-resolution views of fluvial, pluvial and coastal flood risk, including different return periods as well as defended and undefended scenarios. This can help provide a more detailed view of individual exposures and identify concentrations of risk across a portfolio.

A clearer view of risk

Flood modeling has evolved, and so has the information available to underwriters. Historical loss data alone cannot capture the full picture of today’s flood exposure, but advanced models should not be accepted at face value either.

Confidence comes from understanding the data, methodology, validation and uncertainty behind the output. That transparency can support more informed pricing, risk selection, accumulation management and renewal decisions.