On 11 March, a massive earthquake and resulting tsunami struck Japan, causing massive destruction to life, property and infrastructure – even altering the lay of the land itself. Now the (re)insurance industry must attempt to get to grips with the scientific implications of these events
Anyone in any doubt that the 11 March Japan earthquake was an extraordinary event should take a look at the physical changes it has wrought on the country it struck.
Leaving aside for a moment the devastating damage to property and lives, the magnitude 9 earthquake that rocked wide swathes of the country and triggered a deadly tsunami has left lasting marks on Japan’s topography.
According to RMS chief risk officer Robert Muir-Wood, the country is now between four and five metres closer to California than it was before. The outer north-east coast of Honshu, the largest of Japan’s four main islands, has sunk by one-and-a-half metres.
“Even around Fukushima, it has sunk by 60cm-70cm so the Fukushima Daiichi nuclear power plant is now about 60cm-70cm lower than it was before,” Muir-Wood says. Some of the flooding seen in photos of the event, he added, was because the areas are now below sea level.
Muir-Wood says the quake’s displacement – the difference between the position of a reference point and a later position – was 30-40 metres. “This is absolutely enormous and pretty much unprecedented in terms of how much movement that was,” he says.
Expect the unexpected
Not only was the earthquake big, but it also caught Japan completely off guard. Both local and international seismologists had only accounted for an earthquake of a maximum magnitude of 8.2 emanating from the relevant part of the Japan Trench, the subduction zone (an area where one tectonic plate can slip under another) between the Pacific and North America tectonic plates.
“It was a genuine unknown, where science was genuinely lacking in understanding the maximum earthquakes that this trench is capable of producing,” says risk modeller AIR Worldwide managing director Milan Simic. “We can say it was almost a failure.”
The problem was that the tsunami defences protecting the affected communities were built on the assumption that earthquake magnitudes would not exceed 8.2 on the relevant part of the Japan Trench, and thus any resulting waves would be smaller and more manageable.
“If it had been an 8.2 magnitude earthquake, it is likely that the tsunami would not have gone into these communities because the tsunami walls would have held,” Muir-Wood says.
The (re)insurance industry is still picking its way through the losses. Firms have a far clearer picture a month on from the event than they did in the first week, and most firms have now released tentative preliminary loss estimates.
Publicly available estimates to date, where Japanese earthquake numbers have been separated out from the rest of the 2011 losses, total roughly $8bn (see table). Some observers are expecting the total reinsurance bill to be higher – at about half of the total insured losses.
Industry-wide insured loss estimates range from $12bn to $30bn. Many (re)insurers are basing their preliminary loss estimates on an industry-wide insured loss of between $25bn and $30bn.
The losses will affect all types of insurance and thus reinsurance business. This runs the gamut from liability, to business interruption, to life – around 13,000 people had been killed by the events at the time of going to press, with a further 14,000 still missing.
Business interruption, and particularly contingent business interruption, is a significant grey area because it is unclear how much was written and to what extent companies will be able to claim under coverage that is there. But the bulk of the losses will be property-related. Of this, the lion’s share will be commercial and industrial property, as residential property is largely not reinsured in the private market.
“The principal international reinsured losses should come from the commercial and industrial earthquake market,” says Guy Carpenter’s global head of business intelligence David Flandro. “Even in that market, we know that earthquake reinsurance penetration is only around 14%-17%. In the earthquake reinsurance market most of the treaties are capped by the event limits, although those limits are typically set at a very high level relative to exposures.”
However, as all reinsurers are at pains to point out, initial loss estimates are subject to change. The radiation from the stricken Fukushima Daiichi power plant is not significant from an insured loss perspective, because nuclear risks are typically excluded from private market reinsurance covers.
However, it is preventing loss adjusters’ access to the affected areas. Concerns about appearing insensitive by sending in adjusters so soon after such a tragic event present another inhibiting factor.
To add to all the loss and disruption so far, Japan continues to be rocked by aftershocks. On 4 April, reinsurance broker Aon Benfield issued a report saying that more than 830 aftershocks had been triggered by the original 11 March event, with 57 of these greater than magnitude 6. Shortly afterwards, on 7 April, a 7.1 magnitude aftershock hit on 7 April, killing three people.
It will be some time before the (re)insurance industry truly understands its losses from the Tohoku earthquake and tsunami. Structural damage caused by earthquake damage can be less visibly obvious than windstorm damage, for example. It can emerge several months later.
It took around 600 days after the 1994 earthquake in Northridge, California before insurers could get a definitive estimate from paid claims. A year on from the Chilean earthquake, and company loss assessments are still changing.
“We don’t expect to have anything like a definitive loss estimate for well over a year, although we should start to get some clarity in around 12 months,” Flandro says. “We will have an initial estimate based on reported insurance and reinsurance losses when first-quarter results are announced, but that will be subject to revision.”
Despite the time it will take to determine losses, the size and nature of the Tohoku earthquake has already resulted in many lessons and potential changes that the industry needs to take on board. GR
Losses, not events
The unprecedented and unexpected size of the Tohoku earthquake has reinforced the message that (re)insurers should manage their books of business by taking into account potential loss amounts from a territory rather than single events.
According to AIR Worldwide managing director Milan Simic, (re)insurers managing their portfolios on the basis of events would have only assumed a maximum earthquake magnitude of 8.2, and so would have been caught out.
However, companies managing around an overall loss amount for Japan would have taken the loss in their stride. As big as the Tohoku earthquake was, Japan is capable of producing much higher losses from different tectonic sources.
An earthquake hitting Tokyo would be a worst-case event for global (re)insurers. But while Tokyo was affected by the shaking from the Tohoku quake emanating from the Japan Trench, the city would more likely be hit by events from the closer Nankai and Sugami troughs.
“If all of the participants are managing to a loss amount, then $20-$30bn from this event would not be impossible to contemplate from the loss point of view,” Simic says.
The severe damage caused to the Fukushima Daiichi power plant, where the earthquake knocked out the power to the fuel rod cooling system, could have been prevented by more careful thought, according to risk modeller RMS chief risk officer Robert Muir-Wood.
He argues that the power went out because the generators were placed at a low level on the site. Once they had gone down, workers could not get at them to resolve the problem. “There is clearly a set of critical issues that will need to be learned for the whole nuclear industry as to how they think through catastrophes holistically,” he says.
“There are a whole series of events that happened consequently, but the operators had probably thought about them separately and not happening as part of the same event.”
A feast of data
A positive outcome from the 11 March earthquake and tsunami is that it has given seismologists – and modellers and (re)insurers by extension – a chance to study what happens during such high-magnitude events and understand exposures better.
“There has not been very detailed evidence of what happens in earthquakes of that size before Chile and now Japan,” says RMS’s Muir-Wood. “The strength of ground-shaking at great distance from the earthquake and what kinds of damage this causes, especially to tall buildings, is something that has been speculated about. There is going to be much more information now.”
AIR Worldwide’s Simic agrees: “This will create a huge volume of important information that a number of scientific disciplines will be carefully studying for years to come.”
More to come?
The severe aftershocks seen after the 11 March quake, in particular the 7.1 magnitude event on 7 April, has made some industry participants realise that aftershocks – rather than the initial earthquake – could at some point become the major event.
“It is an interesting question whether the impacts from an aftershock could be higher than the impacts in the same location of the original event,” says Muir-Wood at RMS. “We certainly get questions from our clients about that.”
The earthquake has also prompted questions about whether the Tohoku earthquake could trigger events in neighbouring seismic and tectonic systems, resulting in an event closer to Tokyo and thus be even more costly.
“One interesting question is whether this earthquake has affected the probability of earthquakes further south along the subduction zone close to the Chiba prefecture,” says AIR Worldwide’s Muir-Wood. “There is quite a lot of debate on that question.”
“There is now a lot of work being looked at in terms of what such a major rupture of this magnitude could do to the neighbouring faults, namely the Sagami and the Nankai trenches, which are much closer to Tokyo and can cause losses of much greater amounts,” adds Simic. “Most senior scientists in the seismological community are looking to this problem now and are trying to find some conclusions as to the risk levels going forward.”
There are also questions about whether the series of heavy quakes seen recently are part of a trend.
“We now have an unprecedented situation where, if you take all the major earthquakes from 1900, the four largest events have happened in the last seven years,” Simic says.
The trouble with models
Because the tsunami resulting from the Tohoku quake is estimated to have caused about one-third of the insured losses, there is likely to be a rise in demand for tsunami models.
AIR Worldwide, for example, currently lacks a tsunami model. It generated the tsunami portion of its insured loss estimate by using topographical information and wave height data, overlaid with AIR’s exposure database.
Muir-Wood says that while RMS does have tsunami models, they are not a standard feature of catastrophe models. “The reason why there hasn’t been a lot of momentum to make it a standard feature of models was partly because, in Japan, tsunami walls had been built to protect communities for the size of magnitudes which were anticipated then, and tsunami didn’t seem to be a very big source of loss in Japan.”
There are also challenges to making tsunami a part of mainstream cat modeling. Air Worldwide’s Simic argues that it would need to be a global effort, because tsunami occur globally. Furthermore, earthquakes are not the only cause of tsunami. Simic points out that a bulk of the European tsunami to date have been caused by submarine landslides. Another source is volcanic eruption. “If we were to do that, we would want to have a comprehensive set of tsunami-creating sources as possible.”
Devising accurate tsunami models could also be scuppered by a lack worldwide of information on bathymetry – the topography of the seabed.