Collaboration needed to meet challenge of new primary perils

For decades, the insurance market has primarily focused on the catastrophic impacts of tropical cyclones and earthquakes, which were traditionally categorised as primary perils. However, recent events like the California wildfires of 2020, Winter Storm Uri in 2021 and the severe convective storm season of 2023 have shifted the landscape.

These often called "secondary perils" include floods, wildfires, hailstorms, and winter storms, have proven their capacity to generate losses on a par with, or even exceeding, those from tropical cyclones and earthquakes.

Factors such as inflation, climate change, urbanisation, population changes, and infrastructure degradation among others have amplified the frequency and severity of these perils, making it clear the term "secondary" no longer captures their true impact.

According to Aon and Guy Carpenter, the 2023 convective storms alone resulted in more than $50bn in insured losses, highlighting the growing importance of these events in the global insurance market. As the industry grapples with these dynamic threats, it is becoming evident these perils are now integral to the evolving risk landscape, necessitating a reclassification as “new primary perils”.

The mounting cost

New primary perils accounted for 67% of natural peril insured losses in 2023 – 11% higher than this century’s average, according to Gallagher Re, with severe convective storm (SCS) accounting for 57% of global insured losses.

SCS is an intense atmospheric disturbance that can cause powerful winds, large hail, heavy rainfall, and tornadoes. Such an event is often short-lived but can hit suddenly and with extreme intensity. While individual SCS events rarely lead to large-scale insured losses, the real challenge lies in their cumulative effect – frequent, smaller-scale events that add up to substantial costs over time. This makes monitoring and planning difficult and can gradually erode a company's profitability.

Another challenge with these new primary perils is their unpredictable and localised nature, coupled with historically lower level of losses. For instance, the hail event in France in 2022, which resulted in a loss of €4.8bn ($5.3bn) according to the French Federation of Insurance, illustrates this issue. While hail is not uncommon in France, the severity of this event was unprecedented, prompting the insurance market to revise its return periods and reassess risk evaluations for such phenomena.

Beyond SCS, flood causes significant insured losses, but those may not be covered by insurers and reinsurers at the same level in all circumstances, often resulting in higher levels of under-insurance or coverage gaps. This can lead to significant disparities between insured and actual losses, especially in regions not typically perceived as high-risk. For example, in the US, the peril of flood is covered by the National Flood Insurance Program for homeowners (capped at $250,000). Flood is also typically sub-limited for commercial coverage in all of the US.

However, outside the US, flood is typically covered under insurance policies and reinsurance treaties. According to Munich Re, global flood losses have increased over the past few decades and reached $320bn in the last five years alone.

Combining the impact of new primary perils meant 2023 was a high insured loss year, despite the lack of a major event such as a landfalling US hurricane.

Why now?

The rise in these new primary peril insured losses is caused by a number of factors, including socio-economic trends such as inflation pushing up the values of properties around the world, price of repairs, population changes, and increased development of areas (urbanisation, land use changes, degradation of infrastructure) affected by storms and floods.

Modelling for these perils is often thought to be less established than for tropical cyclone and earthquake and it has its unique peril-specific challenges. In some cases, full probabilistic catastrophe models are not readily available from leading catastrophe modelling vendors or geographical coverage is not complete. As a result, less sophisticated methods need to be deployed to be able to assess and price the risk, underscoring the nascent state of modelling for some of these perils.

Where catastrophe models do exist, they are always underpinned by the historical record. In the case of SCS, the historical record was created from observations gathered by individual members of the public, leading to significant gaps as many events went unrecorded. While the global population grows and more sophisticated data ingestion methods, such as satellite observations, are becoming available, the records inherently include biases and inconsistencies in reporting – in some cases, even the event definition is ambiguous. This represents the key challenge for accurate modelling of SCS and a much longer and unbiased observational record is required to capture SCS variability.

Wildfire modelling presents its own unique challenges, particularly because of the expanding wildland-urban interface – the area where buildings and natural landscapes meet. As people move closer to natural areas, it significantly increases the likelihood of human-induced ignition where abundant vegetation provides ample fuel for fires. This means fires can also occur in more unexpected places and in more complex environments.

Aftermath of the Marshall wildfire in Colorado Carl Glenn Payne/ZUMA Press Inc/Alamy Stock Photo

Human influence extends further as some of California's worst wildfires have been triggered by power line ignitions, often exacerbated in windy conditions. In response, mitigation efforts such as burying power lines, implementing power outages during hazardous conditions, and enhancing maintenance practices have been introduced. However, these factors add layers of complexity to accurately capture and model the evolving wildfire risk, as they alter and introduce new variables that must be considered in risk assessment.

Ageing infrastructure also plays an important role in exacerbating the impacts of new primary perils. A prime example of this is Winter Storm Uri in February 2021 – the storm delivered a long-lasting arctic wave of freezing temperatures, ice and snow to two thirds of the US. The situation was exacerbated when the Texas national grid, which operates independently of the national grid, could not meet rapid demand surge during the winter storm.

This escalated further as the storm caused pipeline terminals, wells, and gas processing plants to shut down or operate at reduced rates, leading to a shortage in the natural gas supply used to generate approximately 50% of the state’s power. The final insured loss quantum was circa $17bn (according to PCS), with most loss stemming from cumulative business interruption. These types of effects are difficult to foresee and even harder to model thus are not accurately captured by modelling.

Industry’s reaction

The growing impact of these new primary perils raises concerns about insurability. As these extreme weather events become more common, insurers are having to pay out an increased number of claims and larger amounts more frequently. This has had an impact on the industry in several ways.

Some insurance companies are increasing deductibles for certain perils, adding percentage deductibles for tornadoes or hail events, decreasing limits and/or excluding coverage for certain perils and locations. These increased losses are resulting in insurance premiums rising to levels that many individuals and businesses may struggle to afford or make the decision to self-insure, leaving a protection gap of insurance coverage.

Escalating losses through reinsurance programmes also mean reinsurers are increasing attachment points, reducing capacity in both occurrence and aggregate treaties and increasing rates significantly, which has an overall impact on the premiums paid by policyholders and the insurers’ ability to provide cover.

Meeting the challenge

The increasing significance of these new primary perils to the re/insurance industry means complexity of estimating losses, particularly in areas not historically prone to these events makes it harder for insureds to plan financially for insurance costs.

Reinsurers need to deepen their understanding of these perils by challenging assumptions and refining their grasp of exposure to enhance the precision of modelling. At the same time, catastrophe model vendors must improve transparency about the considerations that inform the development of their models and continuously update their data to better understand the risks involved. This collaborative effort is crucial for creating a bespoke view of risk for these perils, enabling more accurate risk assessment and more effective management strategies.

The industry has made significant strides in modelling and using data to inform its decision-making, but a long road ahead remains. There is still substantial under-collection of exposure information and lack of availability of granular loss data. Explicit consideration of external factors that may lead to loss amplification need to be better understood, perhaps through adoption of counterfactual analysis. Enhanced risk modelling and data analytics needs to integrate with these higher frequency perils to enable more advanced predictive models to account for localised risks.

This will pave the way for next generation catastrophe models that can dynamically adjust to changes in risk levels in near real-time, rather than relying solely on a short-term historical data. Better understanding of the risk could create opportunities for more tailored products that address specific new primary peril risks, rather than being bundled in with many other perils and coverages.

Lastly, the industry can encourage greater collaboration among reinsurers, insurers, governments, and academia to share data, research, and insights on new primary perils. A global standard for modelling and managing new primary perils, ensuring consistency and reliability across the industry would be a lofty goal, but valuable to the industry. This collective approach would lead to more robust risk models and better risk management strategies.

Stephen Young is group head of reinsurance and chief executive Bermuda and Dr Aleksandra Borodina is research lead of portfolio analytics at IQUW