Updated catastrophe models give users more control of systemsReprints
Modelers, reinsurers and brokers are updating their catastrophe models, giving users more control of the data in examining potential events and mitigating losses should they occur.
The changes include removing encryption of the underlying assumptions and data, allowing them to be changed to better reflect a specific portfolio of risks.
Users also can input their specific data and, in some cases, add third-party information to achieve a modeled result that will aid businesses' and organizations' efforts to mitigate their specific catastrophe risks.
“I do believe that we're definitely in a transition mode. Actually, I believe there is a paradigm shift going on in the catastrophe modeling world,” said Karen Clark, president and CEO of Karen Clark & Co. in Boston, who pioneered cat modeling in 1987.
Last week, the Boston-based company said its RiskInsight open modeling platform has been updated to include, among other features, “characteristic event” loss estimates for specific periods of time, and that 10 insurers and reinsurers were already using its capabilities.
The catastrophe modeling marketplace has been “extremely dynamic” the past two years, said Adam Podlaha, global head of Impact Forecasting L.L.C., Aon Benfield's London-based catastrophe modeling arm. “Clients have been calling for models to be open rather than encrypted, so that they can really understand how the models work and also customize them.”
In recent weeks, Aon Benfield introduced a European windstorm model to tackle the No. 1 cause of insured losses in Europe, Mr. Podlaha said.
Its European models, which include flooding and earthquakes, are used mainly by insurers and reinsurers, the company said.
Boston-based AIR Worldwide Corp.'s Touchstone open modeling platform was launched last year and some 60 companies are using it, said Chief Operating Officer Bill Churney.
“With Touchstone, clients are encouraged to conduct more, not fewer analyses, while not having to worry about paying per click. Touchstone integrates seamlessly with our client's internal systems and we see companies hosting their modeling applications both on-site and in the cloud,” he said.
In addition to historical data and policy information, the models use scientifically derived assumptions to produce loss estimates for various types of catastrophes.
Knowing the variables is critical for clients, who want “full transparency,” Ms. Clark said. “If you are a CEO or a board member, you want to know what are the assumptions really driving your loss estimates?”
Catastrophe models generally have four components, said Ms. Clark: Event generation, intensity calculation, vulnerability assessment and a financial estimate.
A user interface sometimes constitutes a fifth component, said Andrew Castaldi, Armonk, N.Y.-based head of catastrophe perils in the Americas for Swiss Re Ltd.
Swiss Re has its own catastrophe modeling system, developed more than a decade ago, that includes all Swiss Re data, Mr. Castaldi said. The reinsurer also is used in underwriting.
“Cat models are more open. You can see what's behind them, what's in the background tables and nothing is encrypted. So if you want to see the underlying data which defines the model, you can actually go into the tables and files and see them,” Mr. Podlaha said.
“Customers are diving deeper and deeper into the assumptions and methodology of our models,” said Rodney Griffin, senior vice president for Oakland, Calif.-based Eqecat Inc., which data and analytics provider CoreLogic Inc. acquired last year.
“We are moving to a situation where we're offering configurability to a much greater extent to allow people to tweak the models,” said Mr. Griffin, but users of such open platform technology must use caution.
“There are some companies out there that perhaps don't have the expertise, loss data, or don't have the background to do an adequate job in calibrating or changing the model,” he said, suggesting that such companies contact their broker for help.
Marsh L.L.C. said it has access to two “leading” software models to help clients quantify and manage their risks.
Enhanced software platforms “have aimed to make the best uses of current computational capabilities such as cloud computing, or making the platforms more flexible to its users ... to achieve transparency in the models,” said John Alarcon, London-based executive director at Willis Re, a unit of Willis Group Holdings P.L.C.
Willis licenses major cat models “to evaluate our clients' risk to natural and man-made hazards and to advise on adequate reinsurance solutions,” Mr. Alarcon said.
“You do need to leverage technology to operationalize the data and analytics and deploy them in your organization,” said AIR's Mr. Churney. Over the next five to 10 years, insurers and reinsurers will leverage analytics more and more.
“You're going to have, I think, an explosion of data and analytics that insurers and reinsurers are going to have at their disposal,” said Mr. Churney. “A key component that is also driving the need to "own the risk' is increasing scrutiny from regulatory organizations.”
Ms. Clark agreed. “Regulators really want companies to have their own story as to why they think their risk is the way it is,” she said.
“The regulatory framework is adding an additional burden to primary insurers in particular, that they own and understand their own risk profile,” Mr. Griffin said.
Another major catastrophe modeler, Newark, Calif.-based Risk Management Solutions Inc., could not be reached for comment.