Predictive modelling remains essential, but full implementation is key: Towers Watson

A survey of property and casualty insurance executives in Canada and the United States has found that while predictive modelling remains essential to performance success for personal lines insurers, its implementation remains uneven and inconsistent.

On Thursday, Towers Watson announced the results of the Predictive Modeling Survey, a web-based survey of 52 Canadian & U.S. p&c insurance executives. The survey found that while 92% of personal lines insurers said that predictive modelling remains essential to performance success for risk selection and rating, only 44% of standard small to mid-market commercial lines insurers said so. Furthermore, only 56% of large commercial and specialty lines insurers indicated that it was essential or very important.

“The effectiveness and extent of modelling implementation is a strong indicator of whether these applications are realizing their full value and contributing to insurers’ profitability,” said Brian Stoll, director of p&c practice with Towers Watson, in a statement. “Beyond national carriers, insurers are only starting to understand the potential of a comprehensive program that applies data-rich analytics to a wider range of insurance functions.” [click image below to enlarge]

More than half (57%) of insurers use predictive modeling for underwriting and risk selection, and its implementation is expected to grow by 33 percentage points over the next two years, the survey said. The long-term growth trend for modelling techniques is consistent across all areas of insurers’ business, with carriers planning to use it for fraud identification (36%), evaluation of litigation potential (46%) and loss control (49%).

Over half of respondents that consider their companies data-driven said they use predictive modelling for other functions, compared with just 12% that are not data-driven

In addition, over half (56%) of respondents that consider their companies data-driven said they use predictive modeling for other functions, compared with just 12% that are not data-driven — a critical divide, since skillful collection and analysis of data can help insurers boost profitability through more accurate pricing, increases in operational efficiency, and more effective marketing and sales, noted Stoll in the statement. “Now that almost two-thirds (65%) of insurers have advanced to where they consider their companies data-driven, these companies are more likely to gain a competitive edge in the marketplace, putting the remaining third of carriers at considerable risk.”

The survey found the average interval between updates of insurers’ models was shortest for personal lines: 1.9 years for both homeowners and auto. For other lines, including specialty, commercial and general liability, the average was between 2.3 and 2.6 years.