Last year marked the 20th anniversary of Hurricane Andrew, an event that changed the insurance industry forever. At the time, says Donald Light, director of Celent's (Boston) Americas property and casualty practice, "Risk models and enterprise risk management systems, as we define and use them today, did not exist."
Now those models and systems are a major part of operations, not just at personal-line P&C companies, but across the insurance industry. Though the big data phenomenon tempts insurers with more and more information that should help inform their risk management strategies, often they find that there is exponentially more that they don't know about their risk profiles. This leads to an even greater push to follow the old Boy Scout maxim -- "Be prepared" -- religiously. For some companies, this means shoring up reserves and taking other financial steps in anticipation of catastrophe-related losses.
"Enterprise risk is something you can't control," says Eric Poe, chief operating officer at auto carrier CURE Insurance ($100 million in assets) in Princeton, N.J. "When you couple that unpredictability with exposure, you arrive at an answer that says, 'Protect yourself at all costs.'"
At CURE, that protection this year came in the form of increased reinsurance. After Tropical Storm Irene caused more than $1 billion in insured losses in the state in 2011, Poe says the difficult decision had to be made to devote more of the budget to reinsurance premium. After all, New Jersey -- the state in which CURE primarily does business -- has become something of a ground zero for major weather events, and the models only indicated that things were going to get worse.
"The reinsurance costs are never too favorable to the primary insurer who purchases it," he laments. "As a rule, reinsurers don't usually make bad bets. So when you have traditional roles of executives who evaluate cost vs. benefit, many of them have a difficult time endorsing the exorbitant costs, which only protect you in a very unlikely rare event."
Of course, it's looking like Superstorm Sandy caused up to 10 times the amount of insured losses as Irene in New Jersey. While Poe wouldn't use the term "vindicated" to describe how he feels today about his push to increase reinsurance coverage, he did say he believes that his "opinion on catastrophic reinsurance is much more credible now than ever before." Every model and data source New Jersey insurers have examined recently say that it's likely that more of these types of events will happen in that area. What that means for operations, however, is the necessity of being ready for as bad an event as possible.
"In insurance, we can always use modeling and analyze trends of events, but with catastrophic weather events there are no credible ways to predict events," he says. "Increasing the level of reinsurance is something we will explore at our renewal."
The wide-ranging and shifting information that can come out of catastrophe models is one reason Karen Clark, a consultant and analyst who some credit with starting the catastrophe modeling industry, developed software at her Boston company, Karen Clark & Co., aimed at reducing insurers' reliance on models in developing their catastrophe risk management strategies.
[10 Pillars Of An Effective Insurance ERM Framework]
"The wide swings in the model-generated loss estimates have resulted in shifting information on the relative risk by geography and occupancy type," Clark says. "Because there is very little data supporting the model assumptions, much of this volatility in the model loss estimates is driven by unknowns and not new scientific knowledge."
Her company's latest product, RiskInsight, "includes a stable set of 'defined-probability' meteorological scenarios companies can use to establish consistent underwriting, pricing and risk management strategies that don't have to change every time there is a model update," she adds.
The data on catastrophes and how they might affect P&C insurers might be top of mind and newsworthy in Sandy's wake, but insurers must manage a number of other risks -- financial, legal and operational, for example. To do so it's not enough for insurers to simply acquire more data -- companies require the ability to analyze and glean actionable insight from the data just as with every other insight opportunity that comes from analytics and big data.
The management at BrickStreet Mutual ($253 million in net written premiums), a small Charleston, W. Va., monoline workers' compensation carrier, understood that there were opportunities to build a better technology framework so business users could more easily implement initiatives aimed at better enterprise risk management. Senior VP and CIO Tony Laska worked with business-side users on the selection and implementation of Denver-based Valen Technologies' InsureRight Risk Report to help it more effectively price policies according to their risk.
The goal is for the enterprise to "use the tools that my group manages to help the other groups, like actuarial, make good business decisions," says Laska. His company, as with many other insurers, is now "focused on how you get that data model right so you can use that analysis not just to make your business grow, but also to mitigate risk as it comes in."
BrickStreet ships data via its warehouse to Valen, which runs it through some algorithms and returns an analyzed file that business users can then interpret using Tableau Software's (Seattle) data visualization offering.
Before implementing these software products, "it might have taken us two days from the time we got [business-side] requirements, tested it and ran it," Laska says. Tableau is easy for the business side to use -- it's a drag-and-drop interface -- and it also "does a bunch of analytics, allowing you to go out there and look at different views of data over time, helping us mitigate risks and resolve our pricing and underwriting," he continues.