Building better models

Dan Dick, global head of property analytics for Aon’s Reinsurance Solutions, explains how the business applies insights and data to the property (re)insurance segment to help define each client’s view of risk.

How are Aon’s academic collaborations incorporated into its Impact Forecasting catastrophe modelling suite?

Our 14 collaborations are all critical to help insurers be better informed on evolving risks. For instance, climate models may be incorrectly representing the Pacific response to greenhouse gas-induced warming with trends toward El Niño-like conditions rather than La Niña-like conditions, according to a new Perspective article in PNAS based on research supported in part by a collaboration between Aon and Columbia University. This inaccuracy could mean that projections of severe weather may be underestimated and that climate modellers and adaptation planners should consider a wide range of potential outcomes. The whole idea is to understand how perils are going to change, publicly share that information, and then build event sets hosted within Impact Forecasting that reflect those changes.

What is the scope of the modelling suite, and which perils are gaining greater prominence?

Impact Forecasting has already built a suite of 130 models across 90 countries and 12 perils. As we continue to respond to evolving losses, we are studying a range of perils from major perils such as hurricanes through to so-called ‘secondary’ perils including wildfires and global floods – even though these should not be considered ‘secondary’ – in order to integrate this data and insight into our model. For example, we are now developing a new severe convective storm model for the US. To do this, we have been finding the right academic partnerships to cover these perils. But we don’t just want to stop there. We want to understand the impact of climate change, and not just for typical catastrophe perils, but also for perils such as extreme heat and drought, and ultimately how this might impact the insurance industry. This understanding will help our clients to navigate volatility and build longer-term business resilience.

What types of information are (re)insurers demanding?

A lot of information that has been captured over the years for underwriting purposes is not always appropriate for the specific risk being insured. Instead we need to focus on granular details most relevant for the specific perils. Wind or hail risk, for example, should have data that focuses on the condition of the roof, the shape of the roof and the type of roof covering. These data attributes are fundamentally more important for the peril than the typical information collected on an insurance application.

We have been seeing more carriers looking to third-party technology vendors that are able to marry imagery and AI to provide qualitative and quantitative information about risks. Aon’s Industry Exposure Database is also helping to fill these gaps. Technology will allow underwriters to augment the data they are collecting and make even better decisions around pricing and reinsurance.

How are clients navigating the current conditions?

We think 2023 will go down as, if not the worst, the second-worst severe convective storm season in US history from an insurance standpoint.

We are starting to see losses pop up in areas that were not typical tornado or hail alleys before. Insurer education is going to become critical, because they want to provide a service and be there to help when these events happen. But they need to get a fair price for the risk they are taking while helping homeowners understand what they can do to mitigate and manage that risk.

Dan Dick is global head of property analytics at Aon's Reinsurance Solutions