Summary: The Pacific Northwest is “due” for an 8.0-9.0 earthquake with an attending tsunami.
This risk doesn’t get enough attention in the insurance community.
If you live in Seattle, Vancouver, or anywhere along the coastline in between and down to Oregon, have a good preparation plan…or move!
One of the things that happens when you spend too much time around insurance is I developed over the years a list of cities I will not live due to their cat risk. It’s a very simple principle: I only live once and thus can not diversify away my risk.
You can imagine high on the list is much of California and Florida. New Orleans was on the list even before Katrina. Houston. Memphis. And the one most people forget about – Seattle.
Certainly people are aware of the quake risk in California. If you are involved in insurance, you likely know about the New Madrid quakes and maybe even Charleston. Few people talk about Seattle as a peak cat risk which amazes me because Seattle is the only major metro area in America capable of having a 9.0 quake – and a tsunami!
The History
The main reason Seattle has been ignored as an earthquake peril is “we’ve never had one there”. What people mean to say is we’ve never had one there in recorded history!
About 20 years ago, scientists did some remarkable work to prove that a major earthquake struck Washington state in 1700 at around a 9.0 magnitude. They were able to compile geological evidence to show that a major tsunami happened around that time period based on the destruction of a forest. There were native oral histories of major earthquakes but they lacked precise detail on timing.
Fortunately, Japan has extensive records on the occurrence of tsunamis and there was a mystery in 1700 – a tsunami without an earthquake. By comparing the sedimentary analysis in Washington to the Japanese records, the researchers realized the “missing” Japanese earthquake was no longer missing. It just happened on the other side of the Pacific! Yes, the Washington quake caused a tsunami that reached Japan!
There are many great resources on this discovery but shorter versions can be found here and here. Note, this is not the first time researchers have discovered important ancient earthquakes. The 2011 Japan quake occurred in an area not believed to be at risk, yet research has documented a major quake happened there in 869. There is much geologists can teach us about areas of risk and the related return periods!
Assessing Return Periods
The 1700 quake originated along the Cascadia Subduction Zone. Cascadia has been thoroughly studied and geologists have produced two important conclusions.
First, a Cascadia quake has the potential to be over 9.0 with experts suggesting a 9.2 is possible. A 9.0 is nearly 50X the strength of the 1906 San Francisco quake and a 9.2 is almost 100X stronger! Note, the 2011 Japan quake was a 9.0 and both it and Cascadia produce major tsunamis.
Second, scientists have examined 10,000 years of sediment and found 20 9.0 quakes in the record and about 20 more 8.0 or higher. That suggests a return period of 1 in 500 for a 9.0 and 1 in 250 for an 8+. The time between 9.0s has been as little as 200 years. It has been 319 years since the last event.
However you want to measure it, we are currently well within the confidence interval for when the next event could occur. Every insurance company that operates in the Northwest (as well as western Canada – Vancouver is at major risk as well) should view their Cascadia risk in the same light as their California risk. Unfortunately, nobody seems to do so.
Estimating Damage Potential
The one publicly available resource I have found on the consequences of a Seattle quake was done by the Earthquake Engineering Research Institute in 2005. Instead of modeling a Cascadia quake, the EERI modeled the local Seattle fault and thus just a 6.7 quake. However, the report is very thorough in regards to assessing all the types of exposure at risk.
Their estimate of economic losses, in 2004 $, is $33B. In 2004, Amazon was mostly a book store. Think of all the growth the area has seen in the last 15 years. If exposure has grown ~5%/yr, then the Rule of 72 suggests exposure since 2004 would have doubled to $66B. A reasonable guesstimate of insured to economic damages suggests there would be $20B of insured damage.
This would be very similar to a 2013 RMS report on a 7.0 on the Hayward Fault in San Francisco causing $20B of insured loss as well as the $15B from the 2011 New Zealand 7.1. But again, the Cascadia risk isn’t a 6.7 or a 7.0. It’s a 8 or 9.0! We are looking at far more than $20B of insured loss.
Seattle vs. Sendai
The 2011 Sendai quake was a 9.0. We can compare elements of that quake to a potential 9.0 in Seattle to better assess the damage potential.
Sendai had a population of roughly one million in 2011. The Sea-Tac metro area is approaching four million. 138,000 buildings were destroyed vs. the FEMA study suggesting 700,000 damaged buildings in the Northwest. There were 20,000 fatalities and 500,000 people evacuated in Sendai. A 2016 FEMA study suggests a 9.0 Cascadia would displace 915,000 with 15,000 deaths. Economic damage in Sendai has been estimated as high as $360B with $40B insured.
I don’t see how a 9.0 Seattle would be less than Sendai so I’ll put my “official” somewhat informed guesstimate at $40-80B.
If you think that’s too high, RMS’s repeat of the 1906 San Francisco 7.9 was $64B in 2006 $. With inflation, that’s probably approaching $100B now. And before you suggest San Fran is much bigger than Seattle, the SF metro GDP is $500B vs. $350M for Sea-Tac. Remember from earlier a 9.0 is 50X stronger than a 7.9 and you can easily make up the difference in exposure. Throw in the potential for damage to spread down to Oregon or up to Vancouver and $100B suddenly doesn’t sound so wild.
Geology Isn’t Seattle’s Friend
A few brief things worth noting about the geology of Seattle. Seattle is much like Mexico City in that the city was built in a basin and sits atop soft soil shaped like a bowl above hard rock. The problem is when an earthquake happens the energy is amplified as it passes through the “bowl” causing greater damage. Anyone familiar with past Mexico City quakes will recognize how this feature can cause tremendous damage.
Additionally, the Seattle area is susceptible to landslides, particularly on Mercer Island. Mercer is the Greenwich of Washington State. The median home price on Mercer is $1.8M. The town has its own landslide site where you can check how likely your multi-million dollar home is to slide into the sea! And that’s without quakes! Imagine the mudslide homeowners loss alone from a 9.0!
Do You Know Your Exposure?
If you’re an investor, do you know what your companies’ PMLs are in Seattle? Is this going to be like Katrina or Sandy, where we don’t find out until after that insured losses were as bad or worse than a Florida event?
For insurers, do you treat your Northwest exposure with the same diligence as your California exposure? Is contract wording as tight? Is your data as accurate? Do you have as good information on what properties can withstand an 8+ event? Do you sublimit property exposures that come through national covers?
Is someone like Boeing, where nearly all their global production takes place in a handful of facilities in Seattle suburbs, priced cautiously because of the risk that 25 or even 50 in process planes could be severely damaged all at once (not to mention the BI risk on shutting down production) or is it dirt cheap because it’s a desirable client?
How about the cloud exposure? No, not to the rain in Seattle, but cloud computing. Amazon and Microsoft, both based in Seattle, are the top two players in cloud services with approximately 50% market share. Sure, the server farms are spread out geographically, but is the infrastructure that ties them together still housed in Seattle? I don’t know. Do you? Is there risk of a severe contingent BI event?
Events That Haven’t Happened Recently Are Always Worse Than the Models Predict
This could probably be an entirely separate post (and some will chuckle that the models are always too low) but the models get closest to the pin on events with historical analogues, e.g. Florida hurricanes. They do worst on places with known exposure but without much data such as New Orleans hurricane or New Zealand earthquake. Seattle doesn’t have any data. It’s been 300 years! The model is going to miss a lot of stuff. If the catalog has a $50B scenario for San Francisco and a $50B scenario for Seattle, I’d bet if they both actually happen the Seattle loss ends up the bigger of the two for this reason.
Conclusion
Don’t move to Seattle!!!
For anyone interested in learning more, there are lots of good links I embedded above and many more I didn’t get to use, so feel free to ask if you’re looking for more information.