Asia’s use of catastrophe models is increasingly sophisticated, but the industry still has some way to go to make the most of the concept

The increasing resolution of catastrophe models signals an important shift in the approach to catastrophe risk management in Asia.

Despite Asia’s disproportionate exposure to global catastrophes, the application of catastrophe modelling is patchy and prone to uncertainty in comparison with more established insurance markets, such as Europe and the USA. However, recent efforts by the catastrophe modelling agencies, brokers, global reinsurers and other stakeholders are enabling insurers to better measure, model and mitigate their exposures.

Among the steps forward was the announcement last year by CRESTA [Catastrophe Risk Evaluation and Standardising Target Accumulations] that it was creating new and more specific zones for China, increasing the number of zones from 60 to more than 2,400. Nevertheless, significant issues in relation to exposure data remain.

“The resolution of nat cat accumulation from cedents’ submissions is at a provincial level for most cases,” managing director for China at Swiss Re, Robert Wiest, says. “We have also received some submissions in prefecture level, one level below provincial level and the level we have been pushing for. Along with the requirement of the China Insurance Regulatory Commission on better data reporting, as well as increasing risk management conduct by insurers, we believe a more refined data reporting should benefit all stakeholders.”

Catastrophe modelling efforts have so far focused on the major Asian markets, in particular China, India and Japan. The major perils are earthquakes, typhoons and floods, with challenges relating to each. While China has a long and extremely detailed historical record of earthquakes, for example, stretching back almost two millennia, cat risk experts must translate the catalogue before they can apply it to the models.

A tale of typhoons

Modelling typhoon hazard, which typically brings much higher precipitation than Atlantic hurricanes, also requires a bespoke approach.

In July, EQECAT announced it would release a basin-wide Asia typhoon model encompassing risk across the entire western Pacific basin (including Japan, China, Taiwan, South Korea, the Philippines, Thailand and Malaysia).

As individual typhoons can have an impact on multiple countries, it factors in the direct effects of wind, storm surge and typhoon-rainfall-induced flooding, while taking into account local building practices, design and building codes.

Also in July, Aon Benfield’s Impact Forecasting launched its pan-Asian typhoon model covering typhoon-exposed regions of China, Hong Kong, India, the Philippines, South Korea, Taiwan, Thailand and Vietnam. “The model gives you a loss answer that covers the wind and flood aspect of the typhoon,” senior analyst at Aon Benfield Analytics, Dustin Fabbian, says. “This is important in terms of the relevance to the local market.”

He sees signs that local insurers in China are beginning to embrace the modelling techniques. “We try to help our clients as much as possible to understand the issues surrounding catastrophe risk.”

Improvements in catastrophe modelling have gone hand in hand with increasing exposures. “The typical approach to assessing catastrophe exposure in the past was for (re)insurers to apply standard actuarial techniques and approaches from traditional markets on historical losses, of which the resulting estimates may be subjected to scrutiny for this region,” Asia catastrophe pool practice leader at Asia Capital Reinsurance Group, Werner Bugl, says.

“In recent years, we have witnessed huge economic and (re)insured losses as a result of catastrophe events,” he continues. “Each of the past three decades has seen new records, both in terms of frequency and severity of events, as well as economic and (re)insured losses.”

In the period 1980-2009, Asia experienced the greatest number of catastrophes in comparison to other continents. Its share of natural catastrophes was at 32%, according to Munich Re, but it only accounted for 9% of insured losses. Many of the catastrophes in the last decade – including the Boxing Day tsunami of 2004 and the 2008 Sichuan earthquake have had high death tolls (190,000 and 70,000, respectively). The majority of Asian countries are moderately or inadequately insured.“

We see a clear trend that the exposure-based risk modelling approach is gaining popularity,” says head of non-life underwriting at Munich Re, Beijing, Fan Weishu. “Gradually, more insurers are starting to accept the concept of risk modelling, rather than sticking with the traditional loss-data-based distribution approach.

“Taking China as an example, the major primary insurers run their nat cat books in vendors’ models through brokers,” he continues.

“Some of them even purchase nat cat models and run them on their own. This trend is encouraging. However, in many countries the modelling applications still lag behind the requirements. The advantage of modelling is far from being taken in full. This is especially true when it comes to data input.”

Data quality

Primary insurers are becoming more sophisticated in their approach to catastrophes, but the market lacks a unified approach. The availability of quality data continues to be a concern in China, notes Lloyd’s in its 2007 report. “Although all of China is now post-coded, the information is not always provided to the insurers, making it very difficult to establish risk accumulations.”

Many insurers continue to collect exposure data at a provincial level, while claims data is often not recorded to geo-coding standards. “Data quality needs to be addressed,” Weishu says. “Besides exposure data, we also need better quality of loss data to validate or calibrate the model output.

“We take it as an encouraging sign that more insurers are consulting Munich Re on data quality, model application and other specifics.” GR