Survey Reveals Predictive Modeling Important to P&C Insurers

Property & casualty (P&C) insurers acknowledged that the capture and transformation of data into useful information has turned into a critical differentiator of performance within the P&C insurance marketplace, according to a new survey by global professional services company Towers Watson.

Executives from both personal and commercial lines insurers participated, with a relatively even divide among small, midsize and large carriers.

“Where there is data, there is opportunity,” said Brian Stoll, director, Property & Casualty practice, Towers Watson. “P&C insurers face difficult long-term competitive challenges, but these can be mitigated for carriers that efficiently integrate predictive modeling and data-driven analytics into operating functions across the enterprise.”

The P&C Insurance Predictive Modeling Survey clearly illustrated the importance of predictive modeling to insurers’ business. Personal lines carriers nearly unanimously (98 percent) said predictive modeling is either essential, or very important, to their business. Eighty percent of small to mid-market commercial lines carriers agreed. Large commercial accounts and specialty lines carriers were less convinced overall, with 55 percent indicating that predictive modeling is essential or very important to their business.

The desire to improve profitability emerged as the leading reason why P&C carriers use predictive models. Ninety percent of all U.S. participants cited a desire to improve bottom-line performance as the primary reason, followed by competitive pressure (75 percent). Larger insurers are actively using predictive modeling in the pursuit of competitive advantage, while smaller carriers have been slower to adopt it.

For example, among personal auto carriers, 92 percent of large respondents use predictive modeling, a number that drops to 76 percent for midsize carriers and 57 percent for small carriers.

The trend also holds true to a lesser extent for standard commercial lines such as commercial auto, where the percentage of respondents that currently use predictive models decreases from 62 percent for large carriers to 6 percent for small carriers.

“Smaller carriers slower to adopt predictive models are at a disadvantage, particularly those competing for business directly against the large insurers,” said Klayton Southwood, senior consultant, Towers Watson. “They face loss of market share and adverse selection as their larger counterparts that have implemented predictive models can target better risks and price more accurately. Smaller insurers need to find ways to follow quickly and leverage predictive modeling despite their relative lack of scale.”

P&C insurers diverge on ways they use predictive modeling for their business. Personal lines carriers are more likely to use models for automated renewal decisions and for pricing than commercial lines insurers. With respect to claim applications, personal lines carriers are more likely to use modeling to detect fraud, while commercial lines carriers use it more to triage claims or to evaluate claims for litigation potential. Dependent variable targeting differs, too, with nearly two-thirds of personal lines carriers modeling on frequency and severity separately, and less than one-quarter modeling on loss ratios.

The survey examined insurers’ most pressing challenges for implementing modeling and ways to improve on techniques. All carriers said they struggle with data and people challenges when incorporating data modeling into rating or underwriting plans, but differences emerged by size of carrier. Nearly three-fourths (73 percent) of large carriers cited both data and IT resource constraints as their top two challenges, while nearly two-thirds (64 percent) of midsize carriers ranked IT resource constraints and over three-fifths (61 percent) listed cultural challenges as their biggest hurdles. Seventy-five percent of small carriers ranked data as their biggest challenge, followed by people challenges (55 percent).

“For large carriers, the sheer volume of customers, producers, data and business lines, combined with cumbersome legacy systems, may explain why data and IT resources present challenges,” said Stoll. “Smaller carriers report data challenges, but don’t voice as much concern about IT resource constraints. This may reflect how, by comparison, the challenge of making better use of their limited data dwarfs other concerns.”

Insurers’ top priorities for improving their modeling techniques differ by line of business. Personal lines carriers intend to spend more time focusing on non-pricing applications, such as development of target marketing lists. Many of these carriers are ready to extend the use of predictive modeling to gain market share in competitive consumer markets. Conversely, over one-quarter of commercial lines respondents indicated that monitoring experience against modeling results will be a higher priority in 2013, in addition to enhancing modeling approaches.

The survey also explored predictive modeling’s impact on insurers’ financial results. Most carriers have improved their bottom-line profitability through predictive model implementation; top-line growth impacts have been less pervasive. Personal lines carriers have realized more benefit from implementation in all measures of top- and bottom-line performance. Midsize and large carriers reported significantly more favorable bottom-line impacts from predictive modeling, particularly in the areas of loss ratio improvement and profitability.

The time elapsed for respondents to actually see financial results varies significantly, with top-line impacts on performance emerging more quickly than those for the bottom line. While most participants said they begin seeing revenue results in a year or less, the time to realize net income results is frequently two years.

“The lead time to realize both top- and bottom-line impacts tends to be much longer in commercial lines. It will be interesting to see whether these times will shorten as carriers continue to feel more comfortable with predictive modeling efforts and become better versed in implementation,” said Southwood.

Towers Watson’s fourth annual Predictive Modeling Survey included executives from U.S. and Canadian P&C insurers. A total of 63 U.S. and nine Canadian executives responded. Responding companies represent a significant share of the U.S. P&C insurance market for both personal lines carriers (17 percent) and commercial lines carriers (21 percent). Respondents were relatively evenly divided among small, midsize and large P&C insurers, as measured by 2011 annual direct written premium, and between carriers that primarily write either personal lines or commercial lines business. Roughly 13 percent of respondents split their business evenly between personal lines and commercial lines. With its range of advanced analytical and modeling software, Towers Watson helps insurers improve pricing performance through predictive modeling.

Source: Towers Watson