Andrew Moore explains how catastrophe modelling works and how he analyses the risk of future storms
“Catastrophe modelling” is a phrase heard more commonly in Bermuda than in many communities around the world. Not surprising, given that this country’s flagship industry is reinsurance, with a particular focus on natural disaster coverage. While many of us are familiar with the phrase, what exactly is “cat” modelling and how does it work?
The definition of cat modelling, according to Verisk, a world leader in the practice, is “using computer programs to mathematically represent the physical characteristics of natural catastrophes, terrorism, pandemics, extreme casualty events, and cyber incidents.” The results enable insurers and reinsurers to evaluate and manage their exposures to specific perils. Another cat model leader, Moody’s RMS, describes the four-module framework under which its models are built. First, the event module, built from a dataset for a specific event, such as a hurricane with a defined strength, size, path and probability of occurrence. Second, the hazard module calculates the strength of the winds around a storm, considering the region’s terrain and built environment. Third, the vulnerability module assesses the degree to which structures, their contents, and other insured properties are likely to be damaged by the hazard. And finally, the financial module, which translates the expected physical damage into monetary loss. Cat modellers, actuaries and underwriters work together, ensuring that the risks reinsurance companies agree to underwrite are priced as accurately as possible.
Andy Moore is one of the scientists working in Bermuda to help underwriters make informed decisions. An atmospheric research analyst at Bermuda-based Arch Re, Moore is a meteorologist who has been working in the insurance industry for 18 years. Moore is passionate, as well as knowledgeable, about the weather. His fascination is derived from his experience sitting on the porch of his childhood home in Ohio with his father watching storms roll in. “There was just something about seeing the lightning and hearing the thunder and seeing a hot summer day turn into this tumultuous, crazy thing,” Moore recalled. After completing a meteorology degree, he took an opportunity to work with Nationwide Insurance, where he became a cat modelling analyst. His career brought him to Bermuda six years ago. He explained what a cat model can do. “You can take a risk and tell the model where the risk is located, even a particular home, and it will tell you the vulnerability. You can then simulate whatever peril you specify against that risk—in Bermuda, a hurricane, for example. It will then tell you the expected losses over a period of time and what a one-in-100-year event would look like.”
For reinsurers, who effectively insure insurance companies, models can be used as a tool to evaluate the risk of an entire portfolio. For example, if an insurance company in Florida covers 50,000 homes, all the information about their location, size, height and age would be fed into the model to come up with a loss estimate for any given year, or in an extreme scenario. Moore is quick to stress the word “estimate,” given the millions of variables involved in such a calculation. “Any model is only as good as the data you feed into it,” he said. “There’s a tremendous amount of uncertainty in each component of the model. Even if you know everything about how often a hurricane hits a certain area, how much damage it actually does to a particular structure is variable.
“Say, you have two houses built next to each other. One of them has its roof fitted on a Monday, after the workers have been partying all weekend—they skimp on a few nails because they’re tired. The other has its roof fitted on a Tuesday, when everyone’s feeling better and they put in the normal number of nails. When you model those houses, the risk is the same, but when the hurricane comes the result is different because of something you did not know.”
Loss experience over decades, or even centuries, can help with the accuracy of models. For some more frequent types of peril, such as thunderstorms, there is plenty of data, but for earthquakes, there is very little. As a reinsurer, Arch Re has access to historical loss data from many insurance companies, perhaps more data than the commercial vendors from whom the company licenses its cat models. One of Moore’s roles is to ensure that the models have as much up-to-date and relevant data as possible and “to take what I know about the science and engineering parts of the models and make sure they make sense for what we’re trying to model.” Risk exposures change over time, with development. For example, coastal areas of the United States continue to see rapid development and the models must keep pace. Climate change is more difficult to build in.
“With Bermuda hurricane risk, for example, we have a historical record that goes more than 150 years,” Moore said. “But the climate is not stationary and all that historical data doesn’t really fully capture what the risk is today. The challenge is how do we take that 150 years and adjust it to where we are currently?” Since 2017, there has been a notable uptick in Atlantic hurricane activity. With every event, much is learnt and new data becomes available to strengthen models and so, effectively, adjustments are being made for climate change with loss experience.
Given that most reinsurance contracts are written on an annual basis, there is not an urgent need for reinsurers to be able to project what climate change will mean to risks 10 or 30 years ahead. However, as Moore observed: “As a reinsurance company, we want to be ahead of the curve and understand where we are and where we might be going, as the Earth keeps warming. That’s one of the major challenges our industry faces today.”
For information regarding how to prepare for an impending storm, contact BF&M.
Read more from our Hurricane Season 2023 series, HERE!