Last Updated 4/12/2022
Issue: A catastrophe model (or “cat” model) is a computerized process that simulates potential catastrophic events and estimates the amount of loss due to the events. Catastrophe models are based on property and casualty lines of business, though property catastrophe models are more widely recognized and have been used to assess risk since their introduction in the 1980s. While property catastrophe (PCAT) models are rapidly evolving as extreme weather events and other perils become increasingly prevalent, the use of casualty catastrophe (CCAT) models is relatively recent and they are still in the early stages of development. However, experts predict that within a few decades CCAT modeling will evolve into a level of sophistication comparable to PCAT models.
Background on Casualty Catastrophes
Casualty insurance, also known as liability insurance, protects the insured in case they are responsible for causing property damage or bodily harm/injuries to others. These policies pay people who file claims against the insured or sue them for damages. A CCAT primarily involves litigation at a much larger scale than individual lawsuits. According to modeling analysts from Guy Carpenter, a CCAT is defined as “an event that causes $100 million or more in direct insured losses from all causes to casualty policies (of all types), with one or more policies and insurers impacted.” Typically, these catastrophic financial losses arise from mass torts. Mass torts include a large number of liability claims from many defendants involving multiple insurer defendants alleging damages or harm arising from a common commercial product or event. A familiar example of a mass tort are mesothelioma claims arising from asbestos exposure.
A key component of CCATs and one reason why they are more difficult to predict than PCATs are the latency periods between when a product is released or an event occurs, and when the symptoms of disease or illness appear. Some lung diseases from asbestos exposure have extensive latency periods, sometimes not showing up until 30 or 40 years after initial exposure. Another well-known example is Vietnam veterans’ exposure to the herbicide Agent Orange. During 1961-1971, American troops sprayed more than 20 million gallons of this chemical, containing the deadly toxin dioxin, in Cambodia, Vietnam, and Laos. It wasn’t proven until years later that dioxin caused various cancers and other serious illnesses.
Casualty catastrophes have the capacity to affect almost all lines of business. General and product liability, professional and directors’ and officers’ liability, business income, workers’ compensation, and employer practices lines are all susceptible to CCAT losses. One current example is the emergence of COVID-19 in 2020. Although the full nature of the impact of COVID-19 on insurance losses may not be known for many years, insurers are bracing for increased levels of potential claims in travel, health, life, event, business interruption, directors & officers, employer’s liability, and errors & omission lines. Currently, litigation is gaining traction against employers whose employees contract COVID at work and then infect their families (“Take-home COVID”).
Casualty catastrophe losses can stem from systematic events or a single sudden event. Systematic events occur when there are multiple claims against many policies from one industry, like asbestos. A CCAT resulting from a single event like the Deepwater Horizon oil rig explosion in the Gulf of Mexico (resulting in the BP oil spill) in 2010 amassed more than 100 lawsuits involving over 100,000 plaintiffs. In 2016 the case was settled for a record $20.8 billion, the largest environmental damage settlement in United States history. Though CCATs are low-frequency, high-cost events, they can have devastating effects on businesses, employees, consumers, and the environment.
The chances of CCAT losses are increasing, as companies today face increasing liability risks due to social inflation. Social trends may lead companies to face more lawsuits from consumers willing to litigate for potentially large monetary rewards from juries unsympathetic to big business. Since casualty catastrophes can happen anywhere and arise from human error across almost any casualty line of business, they can have disastrous and long-lasting implications. Modeling firm analysts from Guy Carpenter noted that CCATs are “the most daunting threat that casualty reinsurers face today.”
Historic Casualty Catastrophes and Industry Attention
Numbers show 2 out of the 3 largest historical insurance loss events have been attributed to casualty losses. Asbestos accounts for the second biggest loss behind Hurricane Katrina, resulting in $100 billion (as of 2019) in insured losses. By 2002, 730,000 plaintiffs and 8,400 defendants were involved in the lawsuit, making asbestos the “largest, longest, and most expensive mass tort in U.S. history.” The large financial losses forced many defendants to declare bankruptcy and lay off thousands of employees because of the extensive litigation costs. Environmental issues such as pollution also rank highly in insurance losses, costing the industry approximately $36 billion. Despite these large losses, and the looming threat of the next big CCAT, there is some concern that the insurance industry directs more attention to PCATs rather than CCATs. There are several explanations that may account for this:
- Property catastrophes happen suddenly, garner more press attention, and the events occur over a relatively short lifespan, making estimates of the loss more easily calculated. Conversely, due to their latent nature CCATS emerge gradually, with lawsuits generating appeals that can sometimes take decades to settle. Therefore, the extent of these losses is not known for some time, and as a result, some insurers under-reserve funds. Under-reserving is the chief cause of insolvency for property liability insurers and a common element of casualty catastrophes. This is what happened with asbestos.
- Quantitative data on casualty catastrophes is hard to find. As mentioned above, the length of litigation prolongs the known values of damages that a court may assess, making it difficult for insurers to calculate what their total losses are. Additionally, many settlements are made privately and the parties involved may sign nondisclosure agreements. This adds to the difficulty in ascertaining a final loss amount. Casualty catastrophes are also difficult to track since they do not have an established identification system, unlike PCATs that are issued a serial number by Property Claims Services that allows for data aggregation of insured losses from an event.
- Difficulty in modeling casualty catastrophes. Property catastrophe modelers have an array of data points to analyze when predicting future catastrophes, such as established weather patterns, past historical event data, and defined locations for events. The opposite is true for CCATs, which largely deal with a complex man-made legal institution that changes over time. Whereas each mass tort is unique, weather patterns in PCAT models share common features and are more predictable. Additionally, casualty events can happen anywhere, whereas property events such as hurricanes are usually restricted to certain geographical locations. Moreover, the sheer number of potential casualty events that could happen makes CCAT modeling more difficult than PCATs which typically have a finite number of possible occurrences. Predicting the next CCAT is especially tricky since it is likely that the next CCAT event will be due to a peril that is currently unknown. All these factors introduce a level of complexity in casualty modeling, but as the history of PCAT models has indicated, these challenges are not insurmountable.
Casualty Catastrophe Models and Datasets Today
Today, CCAT risk is recognized as problematic for the insurance industry. Historically, insufficient casualty reserving has contributed to previous insurer insolvencies. An A.M. Best contributor estimates insurers are on a similar path to financial crisis today after underpricing casualty risk for a decade. One method insurers use to predict future risk so they can price premiums adequately are models. Even though regulations require insurers to quantitatively identify risks from both property and casualty catastrophes the use of models to do so is not mandated. For those companies who choose to use a CCAT model to assess casualty risk, various proprietary models and datasets exist. Companies such as Willis Towers Watson/Willis Re, Praedicat, and AIR utilize data in different ways to assess risk for casualty catastrophes.
Casualty catastrophe models are structurally similar to PCAT models. Both models include modules for event, hazard/intensity, vulnerability, and financial losses. However, even though the components are similar, CCAT models differ from PCAT models in their relative size and complexity.
Scenario-based models show the most potential for future exploration since they do not rely entirely on historic data, which may not be wholly predictive of future casualty events. Praedicat takes this forward-looking approach in their CoMeta model. Instead of using historical data in assessing risk, Praedicat’s model examines possible future casualty events by using big data technology to mine peer-reviewed literature. This type of analysis gives insurers the chance to prepare and price risk adequately. In 2021, Praedicat identified isoxaflutole, nanoscale food additives, and microplastics as products of emerging interest. Talc and glyphosate are also products driving casualty risk and are influenced by social inflation.
Historical datasets on casualty losses are available, as well. Established in 2003, Advisen’s Loss Insights (formerly Master Significant Cases & Actions database, or MSCAd) database tracks outcomes of legal cases, actions, or events that have resulted in (or may result in) $1 million or more monetary judgements against corporations. It currently has more than 800,000 records with court judgments of more than $9 trillion. Advisen estimates the database adds an additional 60,000 records each year.
As usage of CCAT models increases, there will be more financial incentive to support these complex, technical databases and data sets. Advances in data science will benefit CCAT models, ultimately leading to improvement and refinement of the models.
NAIC status: The NAIC’s Center for Insurance Research & Policy (CIPR) hosted an event at the 2021 Summer National Meeting on August 17th, Casualty Catastrophe Risk in a Time of Social Inflation: Landscape, Modeling and Action. Regulators, insurance executives, and modeling analysts from Praedicat and AIR discussed the casualty catastrophe emerging risk landscape, tools available to understand it, and actions industry and regulators can take to address it.
Committees Related to This Topic
Casualty Catastrophe (AM Best, Feb. 2020)
Take-Home COVID-19 Claims: Preparing for a Second Wave of COVID Litigation (Risk Management Monitor, Oct. 2020)
Casualty catastrophes and the next asbestos (Conning Commentary, April 2019)
Casualty Catastrophe Analytics: Where we are and where we should be on this critical risk (Variance, 10:2, 2016, pp. 292-311)
Casualty Catastrophe Modeling (Matthew Ball, Yi Jing, and Allan Cohen, Towers Watson, 2013)
What is a casualty catastrophe? (Praedicat, 2013)
Casualty Catastrophes (actuaries.org.uk)