How Disasters Revolutionized CAT Modeling

“Carriers and MGAs use CAT modeling to gauge the risk of an individual account, which determines their likelihood to offer coverage, the price at which they do so, and the limits they offer, as well as the amount of risk it will inject into their overall book of business,” said George Papadakis, client director at Moody’s Risk Management Services.

“Risk managers need this information as well to determine how high their purchased limits should be and how to structure their policies to make their coverage as cost effective as possible. As intermediaries, brokers need access to these tools to advise their clients and negotiate with carriers on their behalf.”

Catastrophes Spurn Evolution

Hurricane Andrew in 1992 changed the insurance industry’s perception of hurricanes and the need to better understand and calculate the catastrophic damage associated with future weather-related events. Improvements in modeling required better understanding of the science of particular perils as well as additional detail on building construction and the resilience or lack of resilience in a severe storm condition.

“Andrew led to the wholesale adoption of cat modeling by carriers for tracking accumulations of possible losses, and not just exposures,” Papadakis said. “This happened at the carrier level and bled down into the brokerages as they needed these same tools to provide relevant advice to their clients. Model vendors also began to develop more peril models, such as for severe convective storm, winter storm, wildfire, flood, and expanded into more international territories.”

Hurricane Katrina in 2005 resulted in more losses due to secondary flooding than due to the wind-generated catastrophe. After the storm surge data was re-evaluated, many hurricane model profiles needed updating.

“Katrina also prompted the development of loss amplification, or the scaling up of economic losses, due to increased localized inflation caused by shortages of labor and materials after major events,” Papadakis said.

CAT models are continuously being evaluated and revised based on current events as well as changes in building materials, building codes, population density changes, and resulting infrastructure changes.

Exceedance Probability and Hurricanes Ian and Harvey

Exceedance Probability (EP), which plays an important role in CAT modeling, is the probability that a certain loss value will be exceeded in a predefined future time period. For example, if you simulate 10,000 years of hurricanes, the highest loss will have a 0.01% chance of being exceeded. The Probable Maximum Loss (PML) is the peak dollar value of the EP curve and represents the percentile of the annual loss distribution in a given time period (1/250 years, 1/500 years, etc.). The Average Annual Loss (AAL) is the average of the entire loss distribution and is represented as the area under the EP curve. The AAL is the estimated annual premium based on the particular catastrophic exposure.

Models are highly adept at measuring very large events such as Hurricane Ian in 2022. The wind event was well predicted and performed as the models expected, building structures performed as the model predicted, and the flood model predictions were accurate. This would suggest the PML exposures and the AAL premium dollars responded as the clients and insurers anticipated.

“Another example is 2017’s Hurricane Harvey,” Papadakis said. “A large, big-box retail client examined their property distribution and values and had it modeled. The high results indicated they needed higher limits for projected hurricane losses. They also had a high risk for Business Interruption (BI) due to the location of some of their distribution centers. By buying additional BI coverage and higher wind limits, they were able to recover more fully and quickly than they would have otherwise.”