From cell phones to miracle cures, recent advances in modelling have made it possible to prepare for the next major casualty catastrophe, say Matthew Ball and Ed Hochberg of Towers Watson.
Hurricane Andrew was the catalyst that prompted the widespread adoption of natural catastrophe modelling by the re/insurance industry from the early 1990s onwards, as insurance companies realised that their reinsurance protection was often far from adequate. Since then other major events such as the Northridge earthquake (1992), World Trade Center (2001), Thailand floods (2011) and the Japan earthquake (2011) have been the catalyst for increased sophistication in natural and man-made property catastrophe modelling and their use (Figure 1).
The Bermuda market was—and continues to be—at the forefront of this modelling evolution. The development of increasingly sophisticated natural catastrophe modelling has allowed property and casualty (P&C) insurers to better measure and manage their exposures. Its expertise in this modelling has contributed to the Bermuda market’s solidifying its position as the major property catastrophe market.
Bermuda also continues to be a strong market for excess casualty insurance and casualty reinsurance dating back to the birth of ACE and XL. Even where it is not currently a regulatory requirement, many casualty re/insurers are already aware that understanding their potential exposures to casualty catastrophes is vital for continued financial integrity and an important consideration within a sophisticated enterprise risk management framework. Natural catastrophes may normally be the primary threat to a property casualty insurer’s short-term solvency, but casualty catastrophes can also pose a risk too serious to ignore.
While the insurance industry and investors have become comfortable with generally accepted risk and pricing models for property catastrophe risk, the prediction of catastrophic casualty claims has lagged behind that of property catastrophe risk. However, when we look at the top 10 most costly insurance events over the last 40 years (Figure 2), this modelling lag is not because of an absence of major casualty catastrophes. Two of the top three most costly insurance events were, in fact, casualty catastrophes-- US asbestos ($75 billion) and pollution ($36 billion).
Casualty catastrophe models have been around for a while to support reserving practices; for example, the industry commonly uses policy event-based loss models for US asbestos, pollution and financial crisis events. However, the industry has yet to integrate casualty catastrophe models adequately for use in underwriting, pricing and risk management. Here we explore some of the reasons for this disconnect and how recently developed models are helping casualty re/insurers better evaluate and understand their exposure to casualty catastrophes. In the longer term, there is potential to create more innovative risk transfer solutions; this could include developing investor interest in cpaital market products specifically to finance casualty catastrophe risk.
What will be the watershed event for catastrophe modelling? Two of the top three most costly insuranec events were casualty catastrophes. So why isn't casualty catastrophe modelling more widespread?
Defining casualty catastrophes
One of the clear issues in managing exposure to casualty catastrophes is the lack of a uniform definition. Where do you start, when there are no typical casualty catastrophes and therefore no precise way to define them? A casualty catastrophe can be caused by a specific type of event or product which can sometimes affect many points in the commerce chain. There can be multiple plaintiffs and defendants, possibly with a class-action status. Casualty catastrophes are often unique and unexpected, more than even natural catastrophes. However, different casualty catastrophes often share a number of the same characteristics or ‘DNA’ which can provide a framework within which to model future casualty catastrophes.
Asbestos is by far the largest casualty catastrophe experienced by the casualty insurance industry, and it resulted in a new awareness of the risks facing the financial integrity of insurers. Other examples of casualty catastrophes include the financial crisis that started in 2007, Chinese drywall, the BP oil spill, product recalls, breast implants, fen-phen anti-obesity treatment and the severe side-effects of certain other pharmaceutical products. In addition, some natural catastrophes can cause casualty catastrophes, such as the oil spills associated with Hurricane Katrina. Some of the potential causes of future casualty catastrophes could be related to cell phones or pharmaceutical recalls.
Measuring the unquantifiable
Sophisticated models have been employed to quantify the impact of natural catastrophes. While the results of these models spark controversy from time to time, the general acceptance of these models by the insurance, reinsurance, rating and regulatory communities has been a great success. So why do casualty catastrophe models remain behind?
First, there are significant differences in opinion over what the next game-changing casualty catastrophe will look like: its cause and size, the number of entities involved and so on. Cancer from cell phone use would probably affect only a handful of large companies—telecommunications companies and their manufacturers. In contrast, lawsuits alleging damages due to climate change could impact virtually any company that has contributed to the emission of greenhouse gases.
In 2008, lawyers involved in the suits against the tobacco industry 10 years earlier applied similar tactics in Kivalina v ExxonMobil Corp, et al. The suit seeks compensation from 23 large energy companies and emitters of greenhouse gases that allegedly misled the public about their contributions to global warming. Although the case was dismissed, it has been appealed to the US Ninth Circuit Court of Appeals.
In a related case, AES v Steadfast determined that insurers are not obliged to defend insureds against acts relating to climate change, at least in the North American state of Virginia. It remains to be seen whether other states will follow suit.
These examples differ from natural catastrophes, where it is generally understood that the next big insurance event will involve a tornado, earthquake, flood or hurricane striking a densely populated city in a developed nation. For natural catastrophes, there are a relatively limited number of perils and locations to consider, and historical cases can be rigorously studied with the aid of meteorologists and engineers. However, it is likely that the next casualty catastrophe will be due to a peril that is currently unknown.
Another key difference is that natural catastrophe events are often known and widely reported immediately. Although the claims sometimes take years to completely settle, estimates of losses from these events made only a few months after the natural catastrophe usually prove to be reasonably close to the actual losses. This is not so with casualty catastrophes. The ultimate financial effects of an unknown casualty catastrophe can remain hidden to an insurer for many years.
A new approach
These obstacles can, however, be overcome. Instead of modelling the physical characteristics of an event, such as location, wind speed, diameter or seismic intensity for a natural catastrophe, the insurance-related characteristics of a casualty catastrophe event can be modelled and made more manageable by allowing insurers to better understand the magnitude and properties of casualty catastrophe claims that may affect their liabilities.
A key step in developing an effective casualty catastrophe modelling process is to gather a database of historical casualty catastrophes, which includes estimated ultimate losses from a comprehensive portfolio of events dating from as far back as the 1950s to the present (Figure 3). Once built, the database can be successfully used to calibrate the various parameters of casualty catastrophe models describing their characteristics, for example, frequency and severity parameters by line of business, allegation or cause, coverage triggered, and the number of claimants and insurers affected. The re/insurer then overlays its specific exposures to produce exceedance probability curves by class of business and in aggregate.
The results from casualty catastrophe modelling can lead to better underwriting—for example, risk selection, pricing for future events,portfolio optimisation, mitigation of exposure through policy terms—and risk management, for example, validation of capital model assumptions, realistic disaster scenarios, and testing of risk mitigation strategies. In addition, the better understanding can assist re/insurers in the development of associated risk transfer strategies.
The future: investor diversification and opportunities for the Bermuda market
As well as the existing uses described above in underwriting and risk management, more sophisticated casualty catastrophe modelling techniques could open the door to a previously inaccessible insurance-linked securities (ILS) market for casualty risks that has the potential to deliver significant returns. As an alternative highyield asset uncorrelated with equity markets and providing portfolio diversification, property catastrophe funds have become increasingly attractive to institutional investors.
Property cat bond issues—driven by investor demand—are booming and investor interest has reached historic highs. This, coupled with insurers exploring innovative ways to transfer catastrophe risk off their balance sheets and unlock vital funding for business growth, has been a major growth area in recent years for the Bermuda market. Casualty cat bonds and other related ILS products have the potential to provide further exciting opportunities for Bermuda in the future.
A more robust modelling of this exposure could make the transfer of this risk to the ILS market more realistic in the future. A new approach has become essential to understand potential causes of future casualty catastrophes. Insurers can now piece together a more robust picture of their potential exposure to future casualty catastrophes and attract investors in search of new opportunities for diversification.
The Bermuda market has been at the forefront in the use of property catastrophe modelling since the early 1990s, which has heavily contributed to its status as the major property catastrophe market across both traditional and ILS products. The opportunity is now there for Bermuda to be at the forefront in the use of casualty catastrophe modelling and solidify its position as a leader in casualty underwriting and product innovation.
Matthew Ball is director and leader of Towers Watson’s risk consulting and software business in Bermuda. He can be contacted at: firstname.lastname@example.org
Ed Hochberg is executive vice president and global product leader of Towers Watson’s risk transfer products and analytics. He can be contacted at: email@example.com
Casualty catastrophes, Towers Watson,