Model Behaviour

Why has catastrophe modelling become so pervasive? Simply put, historical loss records are insufficient.

Models provide a representation of the frequency and severity of complex physical phenomena. And with this type of detail, models allow for the calculation of future expected loss and the quantification of uncertainty.

Quantification has always been top of mind for property and casualty insurers and reinsurers, but is perhaps even more so today given the ever-evolving risk environment and the ever-increasing capacity to collect, analyze and use data.

“The Cat risk business is a growth industry,” Hement Shah, co-founder and chief executive officer of Risk Management Solutions (RMS), said during his keynote address at the recent Canadian Insurance Accountants Association (CIAA) annual conference in San Francisco.

“We are seeing the unprecedented growth of mega-cities. If natural or man-made disasters occur, they have a profound effect,” said Shah, telling the meeting’s 150 delegates that the San Francisco Bay region, for example, is home to 8 million people and more than $2 trillion in property value.

ASSETS TO SYSTEMS

Of course, consideration of potential damage and costs are beyond any one area. “The interconnected world poses a huge risk. Some of the biggest losses in the Tohoku earthquake were not in Japan, they were in global supply chains,” Shah pointed out. As a vital supplier of parts and equipment for things like computers, electronics and automobiles, Japan’s inability to deliver to global customers was crippling.

Shah called it a “paradigm shift” from Assets@Risk to Systems@Risk. “This can mean everything from global supply chains and logistics to networks and digitalization, to cyber risks and IP (Internet protocol) and reputation,” he said.

Devastating events such as Hurricane Katrina in 2005 – among the most costly Cats logged by insurers, and resulting in more than US$150 billion in damages – uncovered the benefits of Cat modelling, but also revealed its flaws.

The lessons learned only helped to make Cat models better. “Models now calibrate storm surges (water damage). Historically, the major peril was wind. There is a host of factors that were previously overlooked, such as economic demand surge, when excess demand leads to price increases in materials and labour,” Shah told attendees.

Lessons were also learned from the almost US$3 billion in claims data for Hurricane Irene in 2011 and Superstorm Sandy a year later in 2012. Providing the backdrop to better manage coastal flood risk, claims data from the two events revealed basement-level property and contents damage contributed to higher losses, especially in business areas.

Today, companies are now able to more accurately quantify the effect on flood loss at the underwriting phase.

IN THE MAINSTREAM

Cat modelling has come a long way in a short period of time, Shah suggested. “Modelling has moved rates, initiated deals and is now firmly in the mainstream. There’s a saying that, ‘All models are wrong,’ but some are useful,” he said.

“It is all data-driven,” Shah noted. “Models reflect the latest scientific and engineering assessments of Cat risk. The technology has allowed us to capture the unique risk characteristics of a region,” he pointed out.

David Crozier, president and chief executive officer of Everest Insurance Company of Canada, said in an interview that “data does not lie. Models are critical to understand and mitigate risk. This exponential increase in data shortens the cycle time on products and gives insurers a handle on any accumulations.”

ON PACE

The sophistication of models can bolster insurers to extend coverage, not just avoid risk. “The fact is, most of world’s economy is uninsured. Models give useful information that can keep the industry more relevant. When a business is insured and suffers a loss, it is way more resilient and recovery is quickest,” Shah said. “Insurance monetizes risk within the economy and via market forces, and reduces risk over time by creating incentives to mitigate risk (loss prevention),” he explained.

Despite the benefits, insurance as a share of global gross domestic product (GDP) is not keeping pace. The global industry experienced the highest insured catastrophic losses on record in 2011, but the claims paid were less than 30% of the underlying losses.

That is alarming to Shah, who said he feels the global p&c industry has room to “triple its value – but that can only be reached by innovation and pushing through obstacles, many of them systemic.”

Rather than just expanding coverage, Crozier said, the focus should be on tailoring coverages that are more meaningful to clients. “Individuals or groups that share the same risks can benefit from coverages that speak directly to them. Credible data can make that happen,” he suggested.

TAKE THE LEAD

Shah noted that insurance could play a role in “depoliticizing climate change,” pointing out that modelling can calm a volatile issue by providing impartial data.

As an example, he suggested considering hurricanes, where the main peril historically has been wind damage. Losses attributed to wind were typically 80% versus 20% to water damage.

As sea levels rise as a result of climate change, water damage during a storm surge in certain coastal areas now model at 50%. That is something underwriters and reinsurers need to know.

Beyond catastrophes, where does the hot button of cyber risk fall into risk modelling? Regarding cyber risk as part of the new economy that must be serviced, Shah noted, nonetheless, that the risks are not well-understood, making it tough to price cyber.

“The data is limited as companies do not want to publicize the security breaches. And if they do, the loss is not disclosed. When the market is better able to comprehend cyber risks, model them and manage accumulations, cyber could really take off,” he suggested.

Crozier concurred. “How do you manage the contagion when a breach can spread from insurance to reinsurance, to your investment portfolio, your share price, to your reputation? It can morph in so many ways that putting a dollar figure to it is a challenge,” he said.

“Models are critical,” Crozier argued. “When a risk modeller comes up with the ‘absolute worst-case scenario’ and ‘the best-case scenario’ for cyber, the market can say, ‘that falls within my risk appetite’ and start to write business. Cyber risk is so vast that the axiom is, ‘You can’t write it, if you can’t understand it,'” he said.

“Human beings will be the architects of an inevitable digital catastrophe, staggering in scope since it knows no borders,” Crozier said. “We can model the path and fallout from a windstorm, but human behaviour is unpredictable.”

Advised Shah, “The industry can combine capital and know-how, along with data analytics, to develop new solutions and expanded coverage. It is an increasingly risky and interconnected world and insurers need to turn risk into opportunity.”