On the official start of the 2012 Atlantic Hurricane Season, which runs from June 1 until November 30, we spoke with Karen Clark, president and CEO of Karen Clark & Company, about new approaches to catastrophe-loss estimation and preparation. Clark explains that while catastrophe models and software applications continue to evolve over time, the bigger challenge for insurers is realizing the full value of such tools.

The consensus among forecasters is that the 2012 season should be relatively “normal,” with current projections remaining unfazed by the pre-season storms. The 2011 Atlantic hurricane season produced 20 tropical cyclones, 19 tropical storms, seven hurricanes, and four major hurricanes. It featured a record sequence of weak tropical storms, and Hurricane Irene, a powerful Category 3 storm, was the first hurricane of the season. The season tied 2010, 1995, and 1887 for the third highest number of tropical storms.

We do know, however, that hurricane disasters can occur whether the season is active or relatively quiet. It only takes one hurricane (or tropical storm) to cause a disaster and virtually unimaginable destruction. Therefore, it is imperative for insurers and risk managers to adequately prepare for every hurricane season regardless of seasonal outlook.

Clark's discussion of risk-assessment and management processes for this season and beyond begins on the next page. Please note that for the purposes of this article, CE denotes a “characteristic event.”

Q. In terms of what insurers need to be doing, what are the fundamental requirements for effective catastrophe risk management?

A. After consulting with dozens of insurance and reinsurance companies over the past few years, we've found that insurers would very much like to have risk-management metrics with three fundamental qualities: consistency, transparency, and operational ability. In order to effectively manage catastrophe risk, insurers need to fully understand the risk and how it's being measured for their specific books of business. Insurers would also like a tool that enables them to monitor the effectiveness of their risk-management strategies over time.

Q. What is/has been the alternative to the CE methodology?

A. Catastrophe modeling has been the standard approach for measuring and monitoring catastrophe-loss potential. While the catastrophe models provide valuable information, they are not highly effective risk-management tools. The numbers generated by the models tend to swing widely from model to model and update to update, and the opaqueness of the models makes it very difficult for the modelers and the model users to decipher the true drivers of changes in the modeled loss estimates, particularly for company-specific books of business.

Q. What differentiates the CE approach from other methods?

A. The CE approach is transparent and is the right balance between fully probabilistic and deterministic approaches to catastrophe-loss estimation. Many companies, realizing the shortcomings in the probabilistic models, have turned back to scenario-based deterministic approaches that are more concrete but don't give a complete picture of catastrophe-loss potential. CEs are defined-probability events created for the return periods of most interest to insurers—such as one-in-100 and one-in-250 years—and are floated across a book of property exposures to provide a complete analysis of the loss potential from representative return-period events.

Q. How does this complement the catastrophe models and/or address the limitations of the models?

A. CEs are based on the same scientific data underlying the catastrophe models, but instead of generating a lot of random events, the science is used to develop return-period events representative of specific regions and perils. The models generate a large catalog of random events, calculate the losses from each event, and then sort the losses from most to least severe to estimate the 100- and 250-year probable maximum losses (PMLs).

Because of the modeling process, PMLs are not operational and are highly volatile numbers. Instead of one number, CEs provide a range of loss estimates for the 100- and 250-year events that are stable, operational-risk metrics that can be drilled down to counties, ZIP codes, and even individual policies for risk-management purposes. In this way, CEs provide valuable information that addresses the model limitations and complements the model-generated information.

Q. How does it foster a more consistent and transparent view of hurricane risk specifically?

A. The CE footprints are completely transparent to the user so they can be easily peer-reviewed by internal and external experts. The damage functions that are used to calculate the losses are also visible to the user. Because they are based on the most credible and reliable scientific data, the CEs remain constant from year to year, thereby providing a consistent yardstick for measuring and monitoring risk over time.

Q. What in your opinion is the greatest challenge P&C insurers face in predicting and preparing for hurricanes?

A. By focusing on PMLs to manage risk, insurers frequently are surprised by actual events that cause losses over their PML estimates. While this may be because of “model miss,” the more significant problem is that model-generated PMLs mask the large loss potential from 100-year events making landfall in specific locations. PMLs give a false sense of security, while the CE analysis clearly shows where companies have exposure concentrations that will result in losses well above their 100-year PML loss estimates even from 100-year events.

Q. Would you elaborate on how insurers are using the CE methodology to:

1. Better understand cat risk and exposure concentrations.

Because the CEs are fully transparent to the user, insurers know exactly the types of events to which they are managing their business. CEs make the events and loss potential more vivid and real to senior executives and boards of directors. By floating 100-year CEs across a company's exposures, exposure concentrations that may be missed by a model are clearly identified and spikes in large loss potential are highlighted.

2. Manage and monitor cat risk over time. Could this potentially diminish claims volume/costs in the long run?

Because the CEs are constant from year to year, risk-management strategies designed to reduce the peak exposure concentrations can be monitored over time. This will certainly diminish the claims volume for individual companies when major hurricanes occur. By focusing on where major hurricanes are likely to cause very large losses, a company can make sure they are not overexposed to events that will tax their claims-handling ability and potentially impair their solvency.

3. Establish a scientific benchmark to test the cat models.

The CEs are based on the same science underlying the cat models. Because the loss estimates generated by the different cat models and model updates can diverge significantly, the CE loss estimates help a company determine which model is more credible for their specific books of business. For companies that blend multiple models, the CE analysis can support selection of appropriate model-blending weights.

Q. Some media reports speculate that climate change is causing an increase in the severity and frequency of hurricanes. Are insurers lending credence to this? How does this theory align with proven science?

A. There is no proven science with respect to climate change and hurricane frequency or severity. According to the expert opinion of the Intergovernmental Panel on Climate Change (IPCC), climate change is likely to decrease hurricane frequency and increase hurricane intensity over the next 20 years. Neither of these impacts is evident in the data so far, particularly with respect to land-falling hurricanes. For insurers and claims professionals it just takes one major hurricane striking a populated area to create a very large industry loss—and companies should be prepared for this at all times. Historically, the worst loss given today's property values would be a repeat of the 1926 Miami hurricane, which is projected to cause nearly $100 billion of insured loss if it occurs again.

Q. Any surprises during the 2011 season? What about Irene?

A. Hurricane Irene is a perfect example of the benefit of the CE approach. If Irene had maintained Category 3 intensity and taken the same path—hugging the New Jersey coast and making landfall across western Long Island—the industry losses would have been over $100 billion. Individual company losses would have been higher than the model-generated PMLs and would have gone over the top of many reinsurance programs. The Northeast hurricane CE is a Category 3 storm with landfall points at 10-mile increments starting near Long Beach, N.Y. The CE analysis prepares companies for realistic events that take typical tracks for each region.

Special Reports /

Q&A: Karen Clark Discusses Cat Management as Hurricane Season Arrives

Related Stories

Recommended For You