Over the last 30 years, catastrophe loss modelling (CAT models) has revolutionised the (re)insurance industry’s capacity to assess, price and manage risks of extreme events for property/catastrophe business and has provided a shared common language for risk transfer applications. Maryam Golnaraghi of The Geneva Association takes a look at where CAT modelling is headed.
Today, the value of CAT models has also been recognised across many sectors and users, leading to various derivatives of these models that are supported by development professionals, the financial sector and national to local governments, in making risk-based decisions. CAT models, conditioned on rapidly advancing Earth observations and climate change models, offer the opportunity for (re)insurers and other sectors to assess physical risk of climate change with a forward-looking approach building on the Financial Stability Board’s task force on climate-related disclosure (FSB-TCFD) recommendations.
The expansion of CAT models for the (re)insurance industry has been ‘demand-led.’ With rising demand for CAT risk analytics from (re)insurers as primary drivers, a number of other factors have also contributed to the expansion of CAT models since the 1980s, including:
- Scientific progress on the understanding of natural hazards and their characteristics (meteorological, hydrological, climatological and geological);
- Engineering research and testing relating to impacts of hazards on the built environment;
- Progress with geographic information systems;
- Various government-based initiatives and an increasing number of industry-academic partnerships.
The development of CAT models followed a series of insolvencies linked to a number of major catastrophes in the US and Europe in the 1980s. For the past several decades, CAT models have served the (re)insurance industry well, facilitating a strong risk analysis and management culture as well as portfolio management practices of property risks throughout the industry value chain. Over the years, the (re)insurance industry’s reliance on CAT models has increased to the point that in some jurisdictions the regulators require CAT models to be officially certified for use in the market.
Seven key factors for CAT models
Effective development and utilisation of CAT models requires in-depth understanding of the underpinning assumptions, intended usage and model limitations. There is much more that could be done to extend the value of CAT models for the (re)insurance industry. This requires a collective endeavour across (re)insurers, brokers, and model vendors not only to benefit further from the current framework but also to extend the CAT models’ capabilities. Seven key issues need to be considered.
- Methodologies: traditionally, CAT models have relied on statistical techniques using historical data of physical events
- Standards and interoperability: Lack of model standards and interoperability significantly burdens model users
- Data requirements, hazard, exposure and vulnerability: Limited availability of historical data is a major challenge. Quality of hazard and exposure data fed to models determine the quality of model output
- Regulatory issues: CAT models and/or their users are subject to regulatory control in a handful of jurisdictions around the world
- Model validation: This is a critical and resource-intensive activity for CAT model developers and model users
- Open-framework and open-source versus restricted: Challenges related to these methodologies persist
- Resource requirements: Development and utilisation of CAT models is a multi-disciplinary and resource-intensive process.
Improvements and expansions in CAT modelling could not only benefit existing (re)insurance users but also a wider group of stakeholders from private and public sectors and the international community.
The future of CAT models
The usefulness of CAT models to the (re)insurance industry and wider society could be even further enhanced with new climate modelling and observational capabilities, as well as emerging technologies (eg, supercomputers, cloud sourcing, deep learning, visualisation, engineering and materials science).
It would be beneficial to integrate the latest climate science and modelling, specifically:
- Leveraging observations of earth’s climate system at an unprecedented scale
- Incorporating latest understanding of climatic regimes and interconnectivities in the weather patterns
- Advancements in seamless forecasting, from minutes to decades
- Earth system simulations (‘synthetic data’)
- Nested models within climate change scenarios
Big data, satellite and remote sensing, wearable devices, computational advancements, AI and neural networking techniques, along with predictive analytics, are tools that, as they mature, will undoubtedly be co-opted into the new generation of risk models, which will be developed over the next few years.
Recommendations for the way forward
Further leverage and enhance current CAT modelling methodologies and tools
To some extent it can be said that models make markets. In turn, markets are also needed to stimulate investment in the current commercially-driven CAT model paradigm. We recommend a call for action to (re)insurers, brokers, model vendors, the development community and the public sector in the following areas:
- Extend existing models to address current limitations and gaps, eg, business interruption and contingent business interruption and supply chain modelling, economic demand surge, and loss adjustments expenses
- Drive for interoperability
- All stakeholders should scale-up ambition for global coverage of natural peril models for every country, across high-, middle- and low-income countries
- Set expectations of transparency and uncertainty quantification in model design and limitations, while remaining sensitive to considerations for intellectual property
- Improve risk communication of model outputs and related model uncertainty amongst users
- Agree on and develop a uniform international exposure data standard to enable transparency, comparability and acceptance of results and allow for efficient use of CAT models.
Embed latest climate science in CAT models and explore opportunities for improving modelling
While a highly complex issue, integration of the latest climate science, earth system simulations (‘synthetic data’) and nested models within the global climate models into CAT models could potentially be a game changer to evolve CAT modelling towards a forward-looking approach.
Such enhanced models could be critical for integrating climate physical risk into core business, financial systems and investment applications (linking to FSB-TCFD recommendations). Building on the international scientific cooperation in climate science and modelling, this offers the opportunity to extend the CAT loss model value proposition also to support new climate insurance product offerings, both now and for the future.
Consider ‘models of models’ and embrace a systems-based thinking for development of the next generations of CAT models
The overarching benefit of coupling models would be to understand feedback loops and cascading effects within and across sectors better (eg, the water-energy-food nexus). CAT models, extended to reflect climate-conditioned future scenarios, could provide new insights and support policy, planning and decision-making. A
Ms Maryam Golnaraghi is director extreme events and climate risk research programme at the Geneva Association.