Great Kanto & Today’s Japanese Earthquake Risk

JULY 24, 2020

Hello and welcome to the first in a new series of podcasts looking at the impact natural catastrophes have on the insurance industry. My name is Maiclaire Bolton Smith, and I am the Senior Leader of Research & Content Strategy with CoreLogic.

I’m really excited about this first podcast because I’m joined by David Gregory, Sr. Product Manager for Global Earthquake Products.

David, thank you for joining me today.

David: You’re welcome! I’m happy to be here.

Maiclaire: So, fun fact about the two of us – I am actually a seismologist. And David is currently doing his PhD in seismology, so the two of us are excited to share some insight as we both have a deep passion for helping people understand their earthquake risk! So, today we’re going to talk about Japan. We all know Japan is a region with a history of devastating earthquakes. The 2011 Tohoku earthquake and tsunami are still very sharp on everyone’s memories. But Japan has a long history of damaging earthquakes. One of the most devastating was the magnitude 7.9 Great Kanto earthquake that occurred on September 1, 1923.

The event caused great damage to Tokyo and the Kanto region. If this event were to occur today, is could result in approximately $300B in damage to ‘insurable building stock’, and the economic damage could well exceed $1T.  At the time, it was one of the most destructive natural catastrophes in human record.

So even though Japan had a rich history and understanding of earthquakes prior to this event, the destruction that this caused was unprecedented.

So David, can you dig a bit deeper into this earthquake? Which fault did this earthquake occur on and can you tell us a bit about the seismic risk in Japan?

David: Of course. The 1923 Great Kanto earthquake occurred on the Sagami trough, one of the three subduction zones—this is where the tectonic plates meet—that make up Japan’s overall tectonic setting.  It’s another of the three, the Nankai, that typically gets the most attention due to its more frequent nature and the fact that it has a longer stretch along the south coastline of Honshu, the largest and main island of Japan. And this means it can potentially affect a wider area and larger population centers. The third is the Japan trench. This is the location of the 2011 M9.0 Tohoku earthquake, the fourth largest earthquake on record, and this event has garnered much scientific interest over the past 9 years.

But switching back to the Great Kanto earthquake: the earthquake was 32 times less powerful than Tohoku, but the loss potential today is thought to be around 3 times higher; and again, this is namely due to the proximity of the fault to the high population centers surrounding Tokyo, and the rupture area of the Sagami occurring closer to land.

Maiclaire: So, let’s continue to dive a little deeper into this event. What do we know about the Sagami trough, and why did this event happen?

David: So, activity on the Sagami trough has been quite quiet in recent years, and the Great Kanto earthquake is the last event to have occurred when considering events of minimum M6.0. That’s almost 100 years ago now.

Further in the past, though, provides a rich history of the destructive potential the Sagami, even in art.  The famous Japanese woodblock, Hokusai’s ‘The Great Wave off Kanagawa’, which is the image of the big blue wave, has become an iconic image used worldwide to depict tsunami risk from earthquakes, and is actually based off this geographic region. Interestingly there was not an earthquake in this region in Hokusai’s lifetime. He lived until he was almost 90 years. It is likely that the original scene was supposed to depict a non-Tsunamigenic wave, but if it were a Tsunami wave, it would likely be from an earthquake on the Sagami trough, just like the Great Kanto earthquake.

Compared to the Nankai, the Sagami trough has a lower recurrence rate, and the Beta, or uncertainty, is higher. And essentially what we are saying here is that events like the Great Kanto on the Sagami occur, on average, much less often, but are also less regular in their occurrence, or dare I say, predictable, than those events that occur on the Nankai.

The trough itself starts in the Sagami Bay area, where it meets the east end of the Nankai, and extends around 210 miles or so southeast into the Pacific Ocean.

Events from Sagami are often known as ‘Kanto events’, as they affect Tokyo and the wider Kanto region, an area of high population. This area, according to the CoreLogic Japan Industry Exposure Database (IED), are over—so over 30% of the building stock especially at risk from earthquakes due to their age and their building characteristics.

Maiclaire: Were there any key findings from the event? And can you comment on the losses from this earthquake?

David: Well, the losses of the event were catastrophic, and from our sources, the ground motions were strong enough to flatten buildings in Yokohama, closer to the source of the earthquake.

Liquefaction, that is, when the strength of soil is reduced by ground shaking and solid soils behave like liquids. It’s known that most of the buildings located on the silt areas of the Sagami and Tokyo Bay were destroyed as well as buildings on the river deltas of the Sagami and Sumida rivers, the latter running through Tokyo.  Interestingly, the term liquefaction was first coined only 5 years prior to the earthquake, and it would take another 40-ish years before the seismological community would pay close attention to the damage potential from earthquake-induced-liquefaction.  Given that the areas worst affected have soils that are known to be liquefiable today, it’s clear that there was significant damage from liquefaction.

Mass landslides were also triggered; this particularly affected the agricultural and more rural areas in the wider Kanagawa region.

And along the coast, there were tsunami waves of around 5m in height recorded across the bay.  The peak wave height observed is roughly a fifth of the that experienced from Tohoku. And there are differences in topographies of the two regions, of course, but more important is the much higher building density in the south. That’s also along this coast, which is just as true now as it was in 1923.

The Great Kanto event is particularly known for the fire damage caused also. The earthquake struck during the busy day time where much activity was underway.  There were fires that were widespread and quickly engulfed large parts of the built-up areas of Yokohama and Tokyo. Tokyo, being further away from the epicentre of the event, was sheltered from the strongest ground motions; however, much of the city was destroyed from the ensuing fires. And notably, a typhoon was active in the Pacific that caused strong localised winds, and these are known to have significantly contributed to the extent of the fire spread and therefore damage.

To put the fire losses into perspective, if we cross the Pacific Ocean, back in 1906, the San Francisco earthquake is also infamous for the fire damage it caused. At the time building density of these two regions was vastly different, but estimated properties destroyed for the Great Kanto earthquake are around 9 times higher than that for the San Francisco earthquake.

And lastly, there were lasting repercussions for trade as a result of this event: the ‘Business Interruption’ loss of the time, if you will. Yokohama was a thriving port city of over half a million people where international trade had been opened up just 5 years prior. And most of this business was literally washed away, and it took a number of years to get back to the levels of trade seen prior to this earthquake.

Maiclaire: That’s really interesting, David. You talk about lasting repercussions for trade. What about the insurance industry? Did this event have any repercussions in the modern insurance industry?

David: Well, yes. In fact, the Great Kanto earthquake is the base of the Lloyd’s Realistic Disaster Scenarios, the RDS, for their Japan earthquake scenario.

Also back at the time, ambiguity existed in fire insurance policies, where by the insurer would not be liable to pay out for fire damage from an earthquake; so some, or most, Japanese insurers did not have to pay out, legally speaking, although there was some government pressure to the contrary of course.

There’s another parallel that can be made to the 1906 San Francisco earthquake, where Lloyd’s of London insurers had a substantial stake in fire premium in San Francisco and famously decided to pay-out the fire losses from the San Francisco earthquake event with other insurers following suit.

So, we actually see the Japanese insurance industry making similar pay-outs for the Great Kanto earthquake, albeit these were more limited due to the sheer volume of loss that threatened many insurance companies with liquidation. So the combination of these two earthquakes has contributed to some tightening and standardisation of contract wordings; and you could say that, between these two earthquakes, they have influenced insurance policy definitions we have in place even today.

Maiclaire: And what if this event occurred today? What would losses look like in today’s environment?

David: The cat modeler in me likes these questions the best. Historically, the event caused significant loss from every peril and sub-peril, that is found in earthquake cat models today—with the exception of sprinkler leakage, because of course sprinkler systems didn’t exist in 1923.

It’s likely that there was much ‘redundancy’ in the destruction caused by the Great Kanto earthquake, where buildings fell, representing a total loss, and still they were also burned.  In the models, we of course limit the damage to the insured value to prevent double counting. And as building codes have improved over time, we should expect less loss; though areas located in the soft soils on the coast are likely to see similar heavy to extreme loss patterns, due to the combination of losses from ground shaking on soft soils, liquefaction, tsunami damage.

The largest difference I would expect would be a reduction in fire loss. Japan has made vigorous improvements to fire safety from earthquakes in recent years with strong reinforced piping for gas networks and improved sprinkler systems. I’d expect that when looking at individual buildings, some of the reduction in fire loss would be replaced with damage from sprinkler leakage damage.  However, as this prevents the risk of fires spreading, the overall risk is significantly lower.

CoreLogic estimates that the event would cause an insurance industry loss of $100B or 105T Yen, if it occurred today, or 80% of which is from ground shaking. And it’s also important to note as part of this that around a quarter of the building stock affected is on soils that are identified as susceptible or potentially susceptible to liquefaction; these soils, in addition to causing liquefaction, at the very least amplify ground motions considerably, contributing to larger building damages.

Maiclaire: So, to wrap up our conversation today - most of our listeners are homeowners, businesses, and insurers. What advice do you have for them? What can insurers, businesses and homeowners take away from the lessons of past earthquakes like the Great Kanto event?

David: Thanks Maiclaire. First I’d like to thank you for having me here today. But to answer the question, at the core of it, hazard science, engineering science and cat models, are improving as our understanding of earthquakes improve. Now, no model is perfect, but the fact remains that if you are using a robust model based on up-to-date science, you have access to and can make business decisions based on the best available information in the market. And, you are less likely to get caught out, especially by surprises that we’ve already learned from in the past.

Earthquake risk is real, and with significant ground motions, livelihoods and homes can be destroyed. Insurance acts as a lifeline in this case, and the individual homeowner level, knowledge of your local earthquake risk can help you better prepare your home in the event of an earthquake, and that’s from knowing to keep contents bolted down through to a full seismic retrofit.

Overall, there’s one thing that I think is good to remember is that we will never finish learning, and the science will evolve as future events unfold.

Maiclaire: I love that – we’ll never finish learning, and the science will evolve as future events unfold. I’ve always said one of my favourite things about being a seismologist is that I don’t know everything, and I’m constantly learning. So, thanks so much for joining me today, David! I’ve really enjoyed talking about earthquakes with you today! And thank you all for listening to the first in our new series of podcasts looking at the impacts of natural catastrophes on the insurance industry. For more insights on natural catastrophe events please visit