“A race is being run” to implement predictive analytics in commercial lines, says Jim Marino, a former chief executive for two major property/casualty carriers, now a director in the Philadelphia office of Deloitte Consulting.
“There is a tremendous amount of interest in this whole topic,” Marino says. “Companies late to the market will be at a competitive disadvantage.”
In 2003, Safeco became one of the first carriers to introduce an automated small commercial underwriting model.
That model, along with hard market conditions, contributed to a 12.4 point improvement in Safeco’s combined ratio in the first half of 2004, according to a report by Inductis, a professional services firm based in New Providence, N.J.
Around the same time, according to Inductis, The Hartford began utilizing scoring models for small to mid-sized commercial accounts.
Today, up to a third of property/casualty carriers are now implementing some form of predictive analytics in commercial lines, according to Marino.
“We’ve seen a number of smaller regional carriers become very vigorous in this area,” says Karen Pauli, a senior analyst with the Tower Group, a research and consulting firm based in Needham, Mass.
“The large carriers are really moving ahead with implementation and the mid-market carriers are pedaling to keep up,” Pauli says.
In its reports, the Tower Group cites five principal reasons why commercial carriers have pushed the development of predictive analytics in the past few years:
- Companies are seeking to avoid the hard market-soft market cycle with a more independent approach to pricing.
- Companies fear being left behind in embracing new technologies.
- Companies recognize their proprietary data as a key source of strategic differentiation.
- Companies are focusing on efficiencies more than cost-cutting in operations.
- Companies find empirical, data-driven analytics to be a sound response to regulator concerns over the subjective nature of underwriting and pricing decisions.
Commercial lines predictive models draw on a far-wider range of rating factors and proprietary data than credit-based scoring in personal lines, and thus the commercial models will vary from company to company to a greater degree.
“It’s not going to be like personal lines where everyone can copy what everyone else is doing,” says Dave Otto, a managing director in the San Diego office of EMB America, an international actuarial consulting firm.
Companies may be surprised to learn what factors have been found to be predictive of loss--and to learn that much of that data is already captured in their day-to-day operations.
For example, says Marino, an account’s payment history--what payment plans it uses, how many late payments it has--has been found, along with other factors, to be a predictor of risk.
The distance of a risk from an agent’s office has also been found to correlate with risk, Marino says, perhaps because agents may not know far-off risks as well as those close by.
A company’s ranking within an agency’s book of business is another risk factor, he adds. Apparently, the lower you are on the agency’s totem pole, the less desirable the accounts you’re getting.
“The biggest eye-opener is that companies begin to see data in new ways and see the need to capture more data,” says Otto. Even if they’re not actively implementing predictive analytics, he says, “most companies are now looking into capturing the data needed to do predictive modeling.”
Sources for this article agree that systematic, comprehensive collection of data on existing and new customers is the foundation for implementing predictive analytics.
“There has been little effort in the insurance industry to leverage the potential of customer information or of the data that it possesses,” reads the report from Inductis.
“Data needs to be collected from all business units--finance, marketing, sales, actuarial,” the report adds. “In addition to internal data, external data needs to be integrated as well.”
According to the Tower Group reports, one firm indicated that it selected from 5,500 different data elements, many of them acquired from public sources using the Freedom of Information Act, to develop a predictive model for middle market commercial lines insurance.
Equipped with predictive analytics, commercial carriers can more systematically target accounts as businesses, regardless of class.
“Companies are now developing underwriting standards based on the characteristics of policyholders rather than on classes,” says Otto of EMB America. “The class will always be there, but the class will have less weight in a rate.”
“Some insurers will be using predictive analytics within a rating class while other insurers use predictive analytics instead of rating classes,” says Barry Rabkin, senior research analyst for Financial Insights, Framingham, Mass.
Predictive analytics, therefore, will undoubtedly change competitive dynamics in commercial lines. No longer will a carrier be able to distinguish itself solely by maintaining underwriting expertise in a few classes.
In theory, a carrier that relies strictly on traditional class-based underwriting will find its most overpriced accounts “cherry picked” by competing carriers using predictive analytics. In reality, experts don’t expect class specialists to stand by and let that happen.
Observers say niche writers are often better positioned to benefit from predictive analytics than generalist insurers--provided they have captured sufficient data on their risks in a format that can be readily analyzed.
“Predictive modeling is not at all in conflict with conventional underwriting wisdom, but it allows you to apply better pricing,” says Marino. “Having predictive models is a necessary condition, but not a sufficient one.”
“None of us thinks this is the death knell of class-based commercial underwriting,” says Pauli at the Tower Group. “You end up collecting your class-based knowledge into the predictive model and, eventually, the underwriters’ role changes to that of portfolio managers.
“Is everyone happy with this role change? No,” Pauli says. “The skill set needed for a portfolio manager is different than a traditional underwriter’s, which makes the transition difficult for some.”
Commercial Lines 
