Arriving in the midst of state legislative activity regarding the use of credit scores for underwriting purposes is the first study-from EPIC Actuaries (Bloomington, IL)-to examine a cross-industry sampling of policyholders, credit information and other variables. Although previous research suggested that there is a significant correlation between insurance scores and risk of loss, explains Michael Miller, consulting actuary, EPIC, "the correlations found in the latest study were statistically stronger than we anticipated."
The study-which was based on a sampling of 2.7 million automobiles-was commissioned by the American Insurance Association, Alliance of American Insurers, the National Association of Independent Insurers and the National Association of Mutual Insurance Companies. Up until this point, the industry lacked information based on samples of the industry average, because studies of credit history and loss are usually conducted with an individual insurer's proprietary data.
Automating Credit Scores
As the case for credit scores improves, carriers that are already employing them are implementing technologies to enhance the ways in which credit information is obtained and utilized. Such enhancements are necessary to spur competitive advantage, because "there aren't too many P&C insurers that aren't already using credit scores," observes John Lucker, senior manager, quantitative services, Deloitte & Touche (Hartford). There is, however, some variance in the extent to which these functions are automated. "How technology is used to incorporate a credit score is pretty much a wildcard," relates Lucker.
Providence Washington (Providence, RI) uses credit scores to help rate policies for its personal homeowners and auto lines of business in 16 states, explains CIO Ed Leveille. The carrier orders credit information through ChoicePoint's (Fairfax, VA) Web interface. The rest of the process is conducted manually. Currently, Leveille is working on a Web services initiative that will enable the real-time electronic gathering of credit scores.
Like Providence Washington, the typical insurance company buys credit scores from a single provider of credit information. In order to gain competitive advantage, relates D&T's Lucker, insurers are moving towards purchasing credit information and then developing their own proprietary credit scoring methodology. "This way the credit scoring methodology is not a black box anymore," says Lucker. "An insurer will know what is inside it and how to explain it to regulators and policyholders."
Carriers are also taking steps toward increasing the accuracy of their risk assessment by basing credit scores on information obtained from more than one source. This is done to ensure accuracy mainly because "there is an inconsistency that sometimes exists with an indivual's credit information from vendor to vendor," says Lucker.
Placing an additional burden on technologists working towards the automation of credit history inclusion in the underwriting process are changes in legislation that dictates how and what credit information can be used. At press time, of the 27 states that recently adopted legislation, 13 of them have adopted the National Conference of Insurance Legislators (NCOIL) model act regarding the use of credit information in personal insurance. A key element of the model, which was created in November 2002, includes the notification of insurance applicants that credit information will be used for underwriting.
As states continue to regulate the use of credit information for underwriting, more insurers are considering expanding their use of both credit and non-credit information in predictive models, relates Lucker. In order to streamline the automation of the use of both credit and non-credit information in the underwriting process for business underwritten in multiple states, carriers should consider moving toward enterprise rules engines with inference capabilities, recommends Lucker. This type of paradigm can support the use of matrix models that can be customized by business rules to create a one-purpose application to serve multiple states, he relates. This allows for the flexibility to vary the use of credit information by state, and even by distribution channel or market segment.