Calculating Canada’s Hurricane Risk

June 1 marks the opening of the 2011 North Atlantic hurricane season. Like last year, seasonal predictions call for elevated activity, based in part on warmer-than-average sea surface temperatures. Americans criticized last year’s forecast after passing through the season unscathed. But Canadians were busy cleaning up from Hurricanes Earl and Igor, both of which made landfall as hurricanes in the Atlantic provinces. In February of this year, Risk Management Solutions (RMS) released its first-ever Canada Hurricane Model as a part of the Version 11 model suite, allowing stakeholders to estimate losses resulting from hurricanes and tropical cyclones in Eastern Canada. The model quantifies the risk to property, contents and business interruption from winds associated with hurricanes and tropical storms that form in the Atlantic Basin. Users have the option of running either detailed or aggregated losses.

Of particular interest to Canada is the treatment of transitioning storms. As storms travel northward, they transition from pure tropical cyclones to extra-tropical cyclones. The Atlantic Oceanographic and Meteorological Laboratory defines an extra-tropical cyclone as a storm system that primarily gets its energy from the horizontal temperature contrasts that exist in the atmosphere. Extra-tropical cyclones are low pressure systems with associated cold fronts, warm fronts and occluded fronts.

As purely tropical storms transition to extra-tropical storms, many characteristics of the storm change. The storm grows in size, translational speed increases and the wind field becomes less symmetrical around the storm center, with the strongest winds shifting to the right of the storm track. Nearly all tropical cyclones affecting Canada have gone through or are in the process of transitioning.

Representing this unique behaviour of storms in the northern latitudes is critical to understanding hurricane risk in Canada. Wind risk tends to cover a larger area, but is generally less intense. Models that don’t include transitioning behavior will overestimate risk near the landfall, and underestimate risk elsewhere.

The new hurricane model complements the existing RMS natural hazard model suite for Canada – including earthquake, severe convective storm and winter storm perils – to give companies a more complete view of natural hazard risk across the country. Nationwide, the average annual loss for hurricanes is less than that of winter and severe convective storms. But localized loss due to hurricanes and tropical cyclones can easily dominate discussion in Atlantic Canada. Hurricanes make up more than 60% of the modelled losses for Nova Scotia and Prince Edward Island. For New Brunswick, Newfoundland and Labrador, hurricane risk is close to 40% of the overall natural catastrophe risk. The RMS Hurricane Model also covers Quebec and Ontario, areas of lower, but still quantifiable hurricane risk. For example, in 1983, Hurricane Hazel killed 81 in Ontario and caused an estimated $1.1 billion1 in economic loss, according to the Department of Public Safety and Emergency Preparedness Canada.

Historical Losses

Historically, Eastern Canada has experienced damaging tropical cyclones and transitioning storms on an annual basis. More than three storms per year have affected the area since 2000, according to Canadian Hurricane Centre estimates (please see Figure 1). Major events in 2003 (Hurricane Juan) and 2010 (Hurricane Igor) resulted in $230 million and $200 million in economic losses, respectively. Risk in Halifax – as determined by loss cost, or average annual loss normalized by exposure – exceeds that of Washington D.C., Philadelphia and New York City (please see Figure 2). At the extreme, a direct landfall of a Category 3 hurricane in Nova Scotia could cause upwards of a $6-billion industry loss across Atlantic Canada.

Hurricane risk drives global catastrophe premiums and losses. The Canada Hurricane Model offers companies guidance for pricing of both insurance and reinsurance rates. It also provides them with the tools necessary to evaluate capital adequacy and reinsurance allocation, while providing better insight into uncertainty quantification and correlation of risk among different perils.

The release of the Canada Hurricane Model, along with new models for eastern Mexico, Central America (excluding El Salvador) and Bermuda, complements RMS’ existing hurricane models covering the United States and the Caribbean. Each region’s model uses local data and expertise to customize the hazard and vulnerability components, but the regions are linked by one continuous stochastic event set for the entire North Atlantic hurricane basin. Companies with exposure in multiple countries can run a single, consistent event set and roll up their losses over the entire North Atlantic basin.