How a black swan event in Japan taught the industry how much it doesn’t know—and how much it was getting right.
Aftershocks, a tsunami, liquefaction, damage to complex supply chains, nuclear power plant accidents, blackouts and power surges were just some of the events caused by the Tohoku earthquake in 2011. Total economic losses from the event totalled $200 billion— or 3 percent of Japanese GDP—and the significance was impossible to deny. So, too, was the totally unforeseen nature of the disaster’s domino effect that shook Japan to the core.
According to Forbes the official death toll two years on—including the nearly 3,000 people still missing and presumed dead—is brushing 19,000; 128,931 dwellings were completely destroyed and as of March 2013 more than 300,000 people were still displaced.
Dr Jayanta Guin, senior vice president of research and modelling at AIR Worldwide, said: “The fifth strongest in recorded history, the Tohoku earthquake is redefining what seismologists consider ‘the Big One’ for Japan. The Japan Trench—the source of the magnitude 9 quake—is one of the most seismically active in the world and has along record of strong earthquakes. Nevertheless, based on historical seismicity, scientists did not expect such a powerful earthquake in Japan’s northwest region.”
Maiclaire Bolton, earthquake product manager at EQECAT, reflected on how the Tohoku earthquake and its aftermath took the world by surprise. “The Japanese are arguably the most tsunami and earthquake prepared nation in the world,” she said, “and this event surprised them. From a public education perspective, people knew to run to high ground and get to rooftops. They had seawalls in place. The event overtopped those sea walls. If you look at the defined tsunami hazard zone in Japan, this event went far beyond that as well. The event was so much larger than anyone expected.”
A black swan event
According to William Thompson, regional director at Willis Re, the country is in the top three globally in terms of non-life insurance premium volume, contributing approximately ¥10 trillion ($102 billion) annually. Thompson estimates that 25 percent of Japanese catastrophe reinsurance is handled by Bermuda players.
Thompson said, “A country with that size of premium and very significant insured values—and a country that we know has both earthquake and windstorm risk—is clearly a major purchaser of reinsurance. It’s an enormously important marketplace for the vast majority of reinsurers, and certainly for reinsurers with a significant international exposure. Japan is one of the core territories.”
For a central peril in a core territory, the Tohoku earthquake was severely underestimated, as was the tsunami it spawned. According to Bolton, the National Research Institute for Earth Science Disaster Prevention (NIED) and the Headquarters for Earthquake Research Promotion (HERP)—the agencies charged with building the national seismic hazard maps for Japan—anticipated a maximum magnitude of 8.4 on the Japan Trench, where the Tohoku rupture took place. EQECAT’s models went a little further, to 8.75, but still missed the mark of 9.0.
Guin said, “While the reasons behind HERP’s modelling scheme were not detailed in the published report, it is known that HERP heavily weighted historical seismicity data in developing seismic hazard maps for all trenches surround Japan’s islands. The view of fault segmentation, characteristic earthquake magnitude, recurrence rate, and cascading ruptures were all based mostly—if not exclusively—on historical data.”
Bolton is unsure that any modelling techniques would have hit the nail on the head. As she explained: “In the modelling world events like the Tohoku earthquake are sometimes referred to as ‘black swans’. These represent the unknown unknowns. Nobody in the scientific community believed that this would happen.”
While modern technology discounted the possibility that a disaster of Tohoku’s magnitude could occur, historical records indicated that onealready had. According to Bolton, Japanese historical documents and sediment analysed by paleoseismologists indicate that an earthquake in 869AD generated a tsunami about the same size as the Tohokugenerated wave.
Bolton said, “There was a bit of knowledge about this event, but it wasn’t incorporated into the national seismic hazard maps. One of the key things that we’ve learned is how important it is to incorporate this paleoseismic information.”
"While it is not possible for a catastrophe model to capture perfectly the exact parameters of every future event, a well constructed model will capture the full range of potential losses."
Guin added: “A central lesson Tohoku has taught the research community is that it cannot rely on historical data alone to determine the risk from large damaging earthquakes: a region believed to be weakly coupled based solely on the historical record, may actually be strongly coupled. This lesson, along with the fact that so much destruction was caused by earthquake risk factors that previously had not been modelled, shaped the objectives of AIR researchers and modellers in developing the update to the AIR Earthquake Model for Japan.”
According to Rene Lamer, head of Asia Pacific Property Catastrophe Reinsurance at Endurance, “People are paying a lot more attention to the flood component of loss than they did before. Also, as cat models evolve and become more refined, knowledge about the underlying exposures becomes more important, in both increased resolution and more accurate building characteristics. Additionally, leading companies are placing a greater emphasis on scenario analysis and contemplating various situations, be they one large event or a series of events, modelled or not, that can have a significant impact on one’s portfolio across multiple lines of business.”
Even though they were a robust assessment of the hazard, the national risk maps didn’t go far enough, Bolton says, and that has driven EQECAT to reconsider some of the information filtering down to its analysts. As EQECAT prepares to release its new tsunami and earthquake modelling suite—set to launch at the end of 2013—Bolton says they took the disparity into careful consideration.
“We do base our seismic hazard on the national seismic hazard maps from Japan,” she said, “but from a modelling perspective we did nottake everything verbatim. We had an enhanced view of Japanese seismic hazard in a few key aspects, and that includes both timedependant frequencies and maximum upper-bound magnitude. In retrospect it still wasn’t enough, so we did go a little bit further than the Japanese national maps recommend.”
“Obviously we have a lot to learn from this event,” Bolton added. “One of the biggest lessons is that we still have a lot to learn. We have very few magnitude 9 earthquakes to learn from, so from a scientific perspective and from an insurance and reinsurance perspective we’ve learned a lot from this event. We need to learn to expect the unexpected. Our goal is always to provide the most up-to-date scientific models and have something that the insurance industry can really rely on in terms of an improved and best-assessment view of risk for a country.”
The scientific community is taking lessons from the Tohoku earthquake and broadening them in a bid to ensure that the same mistakes don’t happen twice. Bolton said, “This has made us re-evaluate. We’re looking at revising the maximum magnitude to 9 and greater on all the subduction zone sources across Japan and worldwide. The big thing is that we will always keep learning. Lessons can be learned around the world from this earthquake.
“Beyond Japan I look at what we’ve learned from the Tohoku earthquake and how we can apply those lessons learned in North America, particularly for the Cascadia earthquake. We will always be updating our earthquake models.”
“Modelling capabilities have evolved since 2011,” Thompson said. “Tohoku taught the market and those involved in the assessment of seismic risk a lot. We’ve seen the vendors update their models to reflect their perception of changed seismic risk.”
AIR’s new model, updated in July 2013, implements the new scientific research conducted post-event, as well as incorporating the industry’s first fully probalistic tsunami modelling capability. Guin said, “AIR has also included explicit modelling of liquefaction. Although eclipsed by the tsunami and shake damage, the liquefaction damage produced by the Tohoku quake was severe. Widespread areas of the Kanto Plain, especially around Tokyo Bay and the Tone River, exhibited deformed streets, sidewalks, and train tracks as well as damage to underground pipes.
“AIR’s new liquefaction module incorporates recent studies on the surface geology, geomorphology and shear wave velocity of soils in regions prone to liquefaction and leverages recommendations from the Japanese government and findings from the Christchurch, New Zealand earthquake.”
Using claims data from all lines of business, AIR revealed that wood construction is more vulnerable than had been previously thought. Because a vast majority of residential structures are wood, the insight has had a significant impact on Japan’s risk profile. Guin concluded:“Much like the 2008 Wenchuan earthquake in China, the Tohoku quake has shown that historical records alone, even ones as long as Japan’s, are not sufficient to model the future risk. Although challenging from a risk estimation perspective, it is prudent to account for physically plausible scenarios suggested by paleoseismic, geodetic and other studies. While it is not possible for a catastrophe model to capture perfectly the exact parameters of every future event, a well constructed model will capture the full range of potential losses.”
While the modelling and reinsurance industries have a lot to consider in the aftermath of the Tohoku quake, there were two very valuable lessons on the positive side: the strength of relationships and the fact that the catastrophe products were able to pull their weight, even in the direst of circumstances.
Lamer said, “We had strong relationships with our Japanese clients before the quake and if anything they’re stronger now. Japan and the Japanese people endured a truly catastrophic event; they were appreciative of the industry’s response. At Endurance we offered to pre-fund a lot of our obligations, knowing our cedants would need that immediate support. This event really emphasised the value of our relationships and we felt that being as proactive as possible was the right thing to do for our clients.”
Thompson explained: “A lot of people talk partnership, talk long-term, talk continuity. These are concepts that people knew the Japanese cedants placed great value on. What the Tohoku earthquake proved is that the Japanese market as a whole responded very much in accordance with these principles. Cedants worked with their reinsurers and in the vast majority of cases the trading relationships have emerged very strong.” In addition, the products pulled their weight.
Lamer concluded: “The catastrophe product functioned as it was meant to. The reinsurers in the property catastrophe space are generally well capitalised and well run companies. The primary companies suffered a far greater loss although overall they’ve rebounded extremely well; if anything, Tohoku had less of an effect on the reinsurance market. This was an event that Endurance was well prepared for and handled accordingly.”
Japan, property catastrophe, catastrophe modelling, earthquake, fire, flood, tsunami