A Role for Characteristic Event Methodology in Catastrophe Risk Management
When catastrophe models were first introduced in the late 1980s, they provided insurance and reinsurance companies with an improved methodology to estimate potential catastrophe losses from hurricanes. While the models provide a dramatic improvement over the simplistic rules of thumb used prior to their introduction, they are blunt tools that provide rough estimates of loss potential and not accurate numbers. The models also present challenges to users due to the volatility in the model-generated loss estimates and the lack of transparency around model calculations.
To manage risk effectively, companies require a credible, scientific view of the risk that allows for more consistent decisions from year-to-year and maintains the confidence of external stakeholders, such as regulators and policyholders. A new approach, based on characteristic events (CEs) for each peril region, complements catastrophe models, and it provides a more consistent and effective framework for catastrophe risk management. The CE approach delivers transparent information and operational risk metrics that companies can use to proactively develop risk management strategies and monitor their effectiveness over time.
Characteristic Events
CEs are defined-probability events created for specific peril regions and return periods. They are based on the same science and data underlying the catastrophe models but eliminate the volatility in the loss estimates caused by noise in the hazard component of the models. For each hurricane region, CE wind footprints have been created to represent 100-, 250- and 500-year type events. The footprints are transparent to the user and can be peer-reviewed by external experts.
To estimate losses, CE footprints are “floated” along the coast and superimposed on company exposures at 10-mile increments. Instead of producing a point to estimate 100-year probable maximum losses (PMLs), CE scenario losses enable companies to estimate the ranges of 100-year losses by region and to see where they are most exposed to 100-year type events.
Last year, many companies using the model-generated 1-in-100 year PMLs to manage risk saw increases in their PMLs of 100 percent or more. When models change significantly, companies have to re-underwrite their books of business, buy more reinsurance, and/or change their risk assessment techniques and models. While the most recent model updates have been much more extreme than previous revisions, all model updates are disruptive to business strategies.
CEs act as an antidote, providing a wealth of information to insurers and illuminating data previously obscured using models alone. Rather than simply generating numbers, CEs illustrate the event types that can impact a company. A sophisticated exposure management tool, CEs highlight exposure concentration areas. CE losses can be estimated at a location level so the marginal impact of individual policies can be calculated.
Because event losses are additive, CE losses can be monitored at the corporate level and drilled down to individual business units, lines of business, and policies. By weighting individual scenario losses, the “expected” or mean 100-year CE loss can be estimated and compared to the model-estimated 100-year PMLs.
Because CEs do not change from year to year, they provide a common currency and consistent yardstick for measuring and monitoring risk. Risk management decisions can be supported and more easily explained to internal and external stakeholders.
Companies are using CEs in multiple ways to better understand their catastrophe risk and loss potential. Companies use CEs as independent benchmarks to test vendor model loss estimates. Catastrophe model loss estimates also can be compared to CEs by peril region to determine appropriate model-blending weights.
More Than One Tool Necessary
Catastrophe losses are already the largest component of homeowners premiums and a growing percentage of all property premiums. Given their significance, companies require more than one type of tool to help understand and manage catastrophe losses.
Hurricane Andrew was the first industry wakeup call to the need for scientific approaches to catastrophe risk management; RMS Version 11 was the second wakeup call to the over-reliance on the models.
Estimating catastrophe loss potential is an inherently uncertain undertaking, and no methodology can eliminate the uncertainty. CEs help companies make more informed, transparent, and consistent decisions in light of the uncertainty.