Better Risk Assessments Saved Virgin Islands Commercial Properties 37% Over Two Years
Carambola Beach Resort at Davis Bay
Hurricane Katrina changed the way insurance companies and some corporations build catastrophe models and calculate the potential damage from a big storm, leading to more accurate damage predictions, better risk assessments(using something similar to this Aravo, third party risk management software), and more strategic and cost-effective insurance policy decisions, some risk management experts said.
Catastrophe storm models from when Katrina hit 10 years ago were fairly crude and largely used by insurers to assess portfolio exposure–basically attempting to understand how much risk a company or insurer had in a given geography, wind zone, earthquake zone or flood zone, said Steve Truono, vice president of global risk management and insurance at Starwood Hotels & Resorts.
But those models in many cases were inaccurate and weren’t necessarily developed or tailored to address complications that might arise, he said.
“After Katrina something called loss amplification was introduced into the models and one of the things we learned was demand surge,” where insurers were all looking to buy lumber, drywall, glass and other repair products, Truono said.
Prices were higher because of supply and demand, there was claims inflation for the cost of repairs, and the models didn’t address any of that, he said. “Now the models have been tweaked and today on the whole they are not perfect but are more representative of the losses” associated with big weather events.
The improvements are due, in part, to use of Big Data when devising various models to calculate the potential damage from a storm or other weather-related incident, Truono said. “Models today, because of tweaking and improvements, yield a result that has a much higher confidence factor, a much higher accuracy factor,” he said.
Claude Yoder, head of global analytics at insurance broker Marsh, said insurance companies have made substantial strides in catastrophe modeling since Katrina, particularly in how they collect and assess data from storms.
One change was putting more people on the ground immediately following a storm so they could see firsthand what happened, what kinds of buildings suffered what specific types of damage, and then compare that to what was predicted in the model.
This has allowed insurers to better design and tailor policies and to more accurately price them for the risks associated with a specific region or type of storm, he said. And this lets companies make better decisions about what types of coverage they need, and how much to buy.
“Having the benefit of transparency around what the components are within the decision-making framework is causing [companies] to better determine whether to transfer more risk or less risk, or what kind of insurance to buy,” Yoder said. “The data available to us, we put to work for our clients around risk transfer.”
Adopting these more data-centric models has helped Starwood to save $1 million annually in its risk program, Truono said. While that may not sound like a lot of money for a multi-billion dollar company such as Starwood, he said improved modeling also leads to more accurate, and in many cases lower, damage estimates, which affects insurance costs and how long a property remains disabled.
At one of the company’s properties in Cancun, Mexico, it saw a more than a 70% improvement in its average annual loss because of “better information” used in assessments, such as looking at various physical construction details of the property and making sure the models had the correct information going in so they could produce the correct outputs, Truono said.
Similar improvements were seen for damage estimates at properties in the Virgin Islands. Such assessments were conducted at about 13 locations in the last two years, resulting in an average reduced annual loss of 37%.
“These are big numbers. We can anticipate lower damageability, which also more times than not reduces the time we are out of business, which reduces the loss of income, which reduces the amount of insurance I have to buy,” he said.
More precise data from more accurate modeling also helps companies such as Starwood better determine how much coverage to buy—or is available—from any one insurer—which Truono calls a more qualitative than quantitative approach.
“If I have five underwriters, many of them will have limits on how much risk they can take on one [client] or geographic area. Having a better sense of the total exposed risk enables underwriters to deploy more capacity or more capital, or may enable them to participate on risk” that previously they may have stepped away from and not underwritten at all. “It not only allows us a broader array of insurers, it drives competition. It also can have implications on deductibles, all of which translates to dollars and cents.”
Greater use of data analytics is raising the expectation level within companies to improve their risk management programs, Yoder said.
“You are being asked to justify why certain decisions are being made and what information is being brought to bear on the decision-making process,” he said.
Truono said because executive management and boards are demanding numbers and data to back up recommendations, “enhanced modeling helps us provide more objective third-party points what the risks are…then we can make recommendations around that objective analytical data.”