Canada’s Quake Risks

Achieving accurate insured loss estimates for earthquake-induced damage requires a realistic representation of inherently complex natural phenomena, detailed inventories of insured exposure, building and contents damage calculation and loss allocation along the risk transfer chain. Together these factors quantify what catastrophe modelling terms as “earthquake risk.”

RMS upgraded and extended its earthquake model-ling capabilities throughout the Americas in 2009, establishing a consistent view of risk from Canada to Chile.

In the new Canada model, our view of earthquake risk has changed. One might ask why, since no major events have driven claims analyses initiatives or other reviews. However, recent years have seen significant advancements in global earthquake hazard assessment, including improved predictions of earthquake ground shaking and more refined approaches to modelling seismicity. The prior RMS model was based on the current Geological Survey of Canada (GSC) seismic hazard maps, which predate many of these advances. The GSC is currently updating its maps. In the meantime, RMS has supplemented existing data with new approaches to predicting ground motion and modelling seismicity; more specifically, it has applied techniques and data from the recently released 2008 United States Geological Survey (USGS) hazard maps, as well as its own science and research, to achieve a new view of hazard. These hazard updates link to innovations in modelling of building performance during an earthquake, improving estimates of damage and resulting loss.

The re-evaluation of hazard, performance and damage culminates in a refined view of Canada’s earthquake risk. Updates to each component affect this view; different changes have different effects depending on the region in question. Using industry loss estimates as the measure of risk, the new model indicates risk is higher in the West up to a loss return period of around 1,200 years (or an annual probability of about 0.083%), at which point we see a crossover to the East for more severe events. The impact of these changes will vary depending on the portfolio of properties insured; however, the net effect on industry-level estimates of financial loss due to ground shaking and consequent fires is an increase throughout Canada for essentially all lines of business and at all return periods.

DIFFERENTIATING THE VIEW OF HAZARD

Canada’s most populated urban areas unfortunately tend to overlie zones of moderate to high earthquake hazard. In the West, hazard is domi- nated by the Cascadia Subduction Zone, which is capable of generating devastating earthquakes up to magnitude 9, putting British Columbia’s major cities at risk. Although eastern Canada occupies a relatively stable continental region, damaging, moderate to large magnitude events have occurred near Montreal and Quebec City.

To establish the likelihood and severity of earthquakes in a region, the GSC divides Canada into zones, deriving a uniform level of hazard for each zone based on the average frequency and magnitude of events. The prior RMS model correspondingly assumed that earthquake hazard was uniform across a zone, employing an averaged view of hazard that typically did not target specific seismic sources. The new model refines this view, subdividing each zone into a grid of small cells to localize regions of higher seismicity (and thus higher hazard). Maximum potential magnitudes and frequencies are determined using the Canada earthquake catalog, a detailed historical record of earthquakes dating back to the 17th century.

The USGS developed this “smoothed seismicity” approach to refining hazard assessment, which RMS has applied to all of Canada (and the Americas in their entirety). It has shown to be especially relevant in high-risk, high-exposure areas where localized hazard can greatly affect risk. For example, hazard increases significantly around Quebec City, as the refinements show a concentration of historical events that a uniform, regional approach could not capture. The approach leads to locally significant hazard increases in other areas of Eastern Canada as well, but has a lesser effect in Western Canada because subduction zone hazard is primarily assessed using geological data and is only marginally affected by historical event data.

RE-ASSESSING GROUND SHAKING SEVERITY AND BUILDING RESPONSE

The depth, magnitude and type of earthquake, as well as local site conditions, all affect ground-shaking levels. Based on a thorough review of the latest science, RMS implemented updated ground motion models that assess the severity of shaking based on the regional characteristics of the different types of events that Canada experiences.

Although the view of hazard in Western Canada was not greatly revised by the smoothed seismicity approach, it is significantly affected by these ground motion updates (primarily the more accurate characterization of the Cascadia Subduction Zone).The subduction zone generates shallow earthquakes at depths of 5-10 km, as well as events as deep as 50 km. Results show thatVancouver’s risk from these deep events is greater than initially indicated, leading to an increase in the city’s overall probable maximum loss. But, implementation of new ground motion models specific to Eastern Canada leads to a slight decrease in losses, again attesting to the event-and region-specific nature of the model component updates.

In assessing the impact of ground motion on a property, estimates of building and contents damage depend not only on the severity of ground shaking, but building characteristics as well — including age, the number of stories and construction type. The new model supports the increased differentiation of building inventory for more accurate loss estimations, even if any or all of the building characteristics are unknown. If information is lacking, the model assesses damage by assuming building characteristics based on an average representation of the existing building stock broken down by line of business-by region- by occupancy.

Innovative engineering technology allowed for realistic computer simulations and full-scale experiments of how a building responds to strong shaking, further refining property damage estimates and greatly enhancing our ability to model realistically the building performance for an event of virtually any type or magnitude. These damage assessment updates have a mixed effect on regional losses: they slightly decrease losses for personal lines in the whole country, increase losses for the commercial lines in western Canada and significantly decrease losses in the East.

The total picture of risk also includes a consideration of how losses can escalate due to societal and economic factors after an event. Demand surge and regional infrastructure failures can increase insured losses to property and business interruption coverages in a severe earthquake. And consequent effects such as fire triggered by an earthquake can push losses into policies that would otherwise not cover shake damage. Because these earthquake-related fire calculations are directly related to ground shaking severity, estimates of fire losses have correspondingly increased, especially in eastern locations such as Quebec City.

THE IMPACTS

In the majority of cases, the model updates lead to increases in risk for Canada’s high hazard and densely populated regions, roughly translating into a 25% increase in modelled losses across Canada. In the West, updates to the modelling of the Cascadia Subduction Zone and changes to building response characteristics have the greatest effect on losses, with increases on the order 30-40%. In the East, where model results are most affected by the adoption of the smoothed seismicity approach and changes in building response characteristics, increases in loss range from a few percent to around 15%. These changes reflect estimates for an industry-representative portfolio. The resultant changes in loss for indi vidual client portfolios will vary depending upon such factors such as construction characteristics, geographic spread and policy conditions; this in turn could lead to a much wider potential range of increases or decreases in loss at the account level than observed in the general estimates.

Changes in loss overall reflect the updated, highly detailed approach to earthquake risk assessment. The wealth of information provided by the GSC and innovations in seismic hazard assessment have allowed RMS to achieve greater accuracy and a more realistic assessment of risk for improved risk management — the ultimate goal in catastrophe risk modeling.