Modelling extreme liability events—including school shootings
Scenario-based casualty risk modelling can be used to calculate the commercial liability losses from extreme events, such as mass shootings, and in a variety of venues. Such recent events include the Parkland school shooting, the MGM Grand Hotel Las Vegas massacre, the 2016 Bastille Day attacks in France, and the 2013 Boston Marathon bombings. Most scenarios impact more than one insurance line.
The catastrophe modelling firm AIR Worldwide collaborated with RenaissanceRe as part of a joint effort to enhance the industry’s modelling of long-tail casualty risk. As part of the collaboration, AIR is producing portfolio-specific casualty losses based on its exposure management application, Arium.
Arium’s scenario-based loss assessment framework enables companies to systematically measure portfolio loss potential and exposures to simulated scenarios representing both emerging risks and forward-projected historical events. AIR and RenaissanceRe say they will work closely to further refine the industry’s first fully probabilistic model for extreme liability events.
There are many casualty events that illustrate the need for insurers to better grasp these risks. One, for example, is the rise in school shootings in the US. Insurance companies are now turning to new and more sophisticated predictive modelling to try to determine the level of risk involved with major casualty events, AIR claims. At the same time, changes made by the Florida legislature to improve campus safety might make schools more vulnerable to lawsuits from injured parties.
In the aftermath of the February Parkland school shooting, the Florida legislature passed the Marjory Stoneman Douglas High School Public Safety Act to improve campus safety through best practices. Robin Wilkinson, vice president and managing director of casualty analytics at AIR Worldwide, says the law could impact liability in future similar situations, by setting a new bar for school performance. Failure to comply with new standards might make schools more vulnerable to lawsuits.
The modelling company notes that, traditionally, insurance companies providing liability policies to schools, hotels, and other venues cover more ordinary events. But recent mass shootings and bombings present liabilities never fully contemplated—nor the risk adequately priced. This tool can help the insurance industry better calculate previously difficult and unmanageable risks.
It ultimately helps the insurance policyholders such as schools, restaurants, concert halls, and arenas by ensuring they have adequate commercial liability coverage for these events, claims AIR.
A robust tool
Wilkinson explains that the release represents a robust tool for the industry because of the data and number of scenarios it draws from. She says it is populated with defined events, such as the Deepwater Horizon explosion or the collapse of Enron, D&O-events like the Volkswagen emissions scandal, as well as systemic areas of liability like asbestos or opioids or sports-related concussions.
“The full stochastic version, which is being refined with our development partner, RenaissanceRe, has thousands of credible synthetic liability events,” she says.
She says that the early adopters of the model have been mainly large, international, insurers with complex risk accumulations. But more companies will eventually use models such as this as standard, she believes.
“Casualty risk is something that insurers and reinsurers both need to worry about and, in time, we expect that modelling this sort of risk will become standard practice throughout the insurance value chain,” Wilkinson says.
“At this time, we are seeing this modelling solution being picked up most frequently by large global insurers and reinsurers whose large portfolios of seemingly independent risks might, in fact, be part of an interconnected network of casualty risk.
“While Arium is currently being used primarily to assess risk accumulations across a portfolio rather than for individual risk selection, we expect this application of the tool will become more common as casualty modelling becomes better understood.”
The genesis for this project came some 10 years ago when Arium was commissioned to ‘do something’ to model casualty accumulations as it was recognised that the casualty insurance market was being underserved by the modelling tools that had helped to dramatically transform the natural catastrophe/property insurance market beginning in the late 1980s.
Wilkinson says that Arium is trying to bring this type of analytical risk management to the casualty insurance market. Its models utilise an extensive supply chain network to define the ways that liability can spread from one industry to affect many others and enable re/insurers to better monitor their risk accumulation across all types of business.
The development of Arium represents a breakthrough, especially in the sense that a framework needed to be developed to analyse casualty accumulations.
“We would say that casualty modelling, as well as the need to collect the data necessary for running casualty models, is becoming better understood, and we are seeing increasing adoption throughout the insurance value chain,” she explains.
“The first breakthrough was to find a framework to analyse casualty accumulations, which had traditionally proved to be a challenge given the diversity of events, the potential range of losses, the issues with uncertainty in the long tail, and the fact that many lines of business can be implicated in different events.”
The second challenge was to estimate the risk of individual events, which Arium does to give both annual average losses and PMLs.
“The next step with RenaissanceRe in this ongoing maturity is developing the first probabilistic casualty models that will enable companies to examine their risk to a full range of potential outcomes,” she says.
“We view this as an incredibly important step in the evolution of this market as it has the potential to bring robust risk management into the casualty insurance space. However, the intention is to run the current scenario-based tool in parallel with the full stochastic model as the former gives transparency and validation of individual scenarios, and the ability to dive into the detail and assumptions—and stress test these—down to the policies that may be brought into that scenario, while the full stochastic tool provides EP curves and heat maps.”
Now the model is up and running, interest is growing as more and more companies recognise this as an area where significant risk has not yet been accurately quantified.
“This point is certainly driven home by the major liability events that continue to unfold in the news every day. These types of events span a huge range of scenarios—everything from financial malfeasance, to food contamination and additives and systemic long-term issues such as crop pandemics, climate change related liabilities and liability related to mass shootings—and dramatically illustrate the need for companies to begin to understand and manage the risk in their casualty portfolios,” Wilkinson says.
She also stresses that a lot of work remains to be done. The immediate future lies in extending the current scenario-based casualty risk modelling framework to enable companies to develop a fully probabilistic view of their casualty risk, for example.
“In other words, we are working on developing the tool so that, instead of saying ‘if event X happens, you would lose $500 million’, we can say ‘given your exposures and the universe of possible events in our tool, you have a 1 percent chance of losing $5 billion’.
“Being able to accurately quantify casualty risk in this way will enable companies gain a real understanding of the risks they are underwriting and, ultimately, to make smarter decisions going forward.”