Uncertainty in the science of catastrophe risk, and how this translates into modeling, has always been a hot topic, not least following Risk Management Solutions' updates to its U.S. Hurricane and Europe Windstorm models last year.
Significant new, peer-reviewed research and data led to substantial changes in the view of risk—and these models are the most informed to date. Yet despite these advances, there inevitably remain areas of uncertainty that may lead to future changes as our knowledge and understanding, as well as technology, continue to evolve.
To use a simple analogy: The landscape of catastrophe risk can be likened to that of Mount Everest, with its steep, unstable slopes, safer plateaus and changing weather conditions. Following expeditions such as those of pioneers like George Mallory and Andrew Irvine in the 1920s—or more recently with the launch of satellites—new observations and data have resulted in new maps, with new measurements of mountain heights and more detailed topography of the slopes and valleys.
The mountain hasn't fundamentally changed, but our understanding of it has developed through ongoing exploration of a previously impenetrable landscape.
Similarly, our understanding of catastrophe risk has significantly improved over the past 20 years, with the addition of knowledge from actual catastrophe experience, new research findings and more advanced technology akin to the satellite. The model-development company's role continues to be leading new explorations and improving the maps and equipment used by the industry.
By their nature, catastrophe models are not based on perfect knowledge or experience, and they continue to evolve. Today's insurance market faces the deep challenge that arises from a desire to access as much new information as possible while needing to rapidly adapt to that new information.
At the same time, multiple external market forces are accelerating change in the insurance industry in a broader risk-management context, with regulatory and rating-agency changes driving companies to take more control of their view of risk.
As a result, a number of key questions remain top-agenda items, including: How to maintain control and modify the assumptions in the model; how to compare the models intelligently; and how to understand the bets placed on the model.
Similar to the way different authorities have different estimates of the height of Mount Everest, modeling companies continue to have different points of view on many aspects of catastrophe risk, leading to different model results. The question remains for the industry over how to utilize these different sources of information within effective catastrophe-risk-management practice.
Some sectors of the industry suggest that model-blending is the solution to managing model uncertainty. However, as highlighted in the Association of British Insurers' "Industry Good Practices" guide, published in December 2011, using a simple blend of results from multiple models may increase uncertainty—and this approach can give a false sense of security.
Using the Tohoko Earthquake in 2011 as an example, the consensus scientific view precluded an event of that magnitude happening on the Japan Trench, so no degree of model blending would have accounted for it. Others within the market prefer the option of choosing to run one catastrophe model in-house and understanding that model extremely deeply.
Insurance and reinsurance companies need to develop a deep understanding of the bets they make relative to the models they use; to integrate their own experience and unique characteristics; and to understand the sensitivity of their loss results to key model assumptions.
Stress-testing model sensitivity is a core part of this process and informs strategic decisions such as optimizing to the model in the pursuit of maximum return vs. minimizing risk through a more conservative approach.
Catastrophe-model users in today's risk-management and competitive environment must increasingly take control and engage with the uncertainty inherent in all models directly.
The catastrophe modeler's role is to empower this enterprise through ongoing exploration, transparency and the provision of a platform on which the landscape of risk can be understood more deeply.
As we look to the future, rapidly advancing technologies and paradigm shifts in data, computing power and business intelligence will transform the way companies explore and manage their risks. Like Everest, the catastrophe-risk landscape will continue to change—and the maps and technology used to assess and navigate that landscape will continue to advance.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.