AI-Driven Property Data Reshapes How Insurers Price Hail and Wind Risk
Moody's found that artificial intelligence-derived property data can shift modeled severe convective storm losses by more than 15% at nearly half of all locations across US insurance portfolios. The finding suggests that enriched building attributes-roof condition, cladding type, vegetation, solar equipment-matter far more than traditional underwriting variables alone.
A Moody's white paper examined how secondary data attributes change catastrophe model output for hail and wind events. While the portfolio-level impact appeared modest-a 5% reduction in average annual loss-this headline figure masked major redistributions of risk underneath. More than 226,000 properties saw material increases in modeled loss, while over 266,000 experienced decreases of similar magnitude.
For underwriters and risk managers, the implication is straightforward: a small portfolio credit can conceal a major reshuffling of where losses concentrate. That redistribution matters more than the top-line shift because it changes which properties are mispriced and where accumulation risk actually sits.
Texas Shows How Building History Affects Models
Texas illustrated the regional effect. State-level modeled loss for severe convective storm fell 11% after enrichment, driven largely by recent roof replacements following multiple large hailstorms in Dallas-Fort Worth and surrounding areas. More than two-thirds of sampled Texas properties had roofs in excellent condition and brick veneer cladding, both of which perform better against hail.
Within the same state, however, properties with older roofs and more vulnerable cladding experienced the opposite effect. Enriched data pushed modeled loss higher for these homes. The ability to distinguish systematically between recently upgraded stock and properties where vulnerability persists creates an underwriting edge that standard submission data cannot provide.
Tree Cover: Asset in Hail Zones, Liability in Wind Zones
Dense tree cover produced opposite effects depending on the dominant peril. In hail-prone regions like Texas and the Great Plains, trees shield roofs and exteriors, reducing impact damage. Properties with high tree density saw decreases in modeled loss after enrichment.
In straight-line wind zones such as parts of the Northeast and California, the same trees become a debris hazard. There, enrichment led to increases in modeled loss for tree-dense properties. The same physical feature carries opposite risk depending on geography and sub-peril.
Two Similar Homes, Vastly Different Modeled Risk
Moody's compared two three-story homes in Aurora, Colorado, a high hail-frequency corridor. One recorded a 62% increase in modeled loss after enrichment; the other saw a 23% reduction. The difference lay in secondary attributes, not age.
The higher-risk property had wood cladding, a steep gable roof with more exposed surface area, a roof-mounted solar array, and limited tree shielding. The lower-risk home featured brick veneer, a hip roof, some tree protection, and no rooftop equipment. The riskier house was newer based on year built, yet specific modifiers outweighed age in the enriched run.
Without enriched data, a carrier might treat the risks similarly or even favor the newer build. Enrichment allows pricing the more resilient property more competitively and flagging the other for closer underwriting review.
Implications for Reinsurance and Data Refresh
Reinsurers can use enriched data to benchmark cedant portfolios beyond aggregate metrics. A book skewed toward well-maintained, recently re-roofed homes is not equivalent to one dominated by older, deteriorating stock, even if headline figures match. Consistent secondary attributes across all locations strengthen confidence in cedant submissions and inform treaty structures.
The analysis raises a practical question: how often should exposure data refresh? Roof age, condition, vegetation, and cladding change over time as properties are maintained, upgraded, or neglected. If catastrophe models rely on static data captured at binding and rarely updated, results drift from current risk profiles. Aerial imagery-based enrichment can provide a scalable way to keep secondary modifiers current across large books, improving explainability and stability of model output.
As the US property catastrophe market enters what some view as a softer phase after years of rate hardening, the ability to separate higher- and lower-risk properties within seemingly homogeneous books may determine which insurers grow profitably and which accumulate exposure on less informed terms. AI for Insurance applications like this one are shifting how carriers assess and price individual risks at scale.
Your membership also unlocks:
