Auto Insurance Companies Relying on Predictive Modeling to Change Business Methods
The prevalence and sheer volume of data available to not only individuals, but businesses, governments, and education institutions as well is truly changing the way we live. The insurance industry is certainly included, and recent changes in the way insurance companies use data could forever change the way insurance is handled, particularly when it comes to auto insurance.
A recently published article by the Insurance Journal points not only of the potential impact the information age, but also examines how the field has already changed dramatically over the past 20 years simply by studying existing data on customers and learning how to interpret it from a business standpoint.
The use of credit scores was previously the most influential to emerge as companies began investigating credit scores, and if customers paid bills on time, to determine the risk associated with insuring a specific individual. Studies showed that individuals who were frequently late in making payments were more likely to file an auto claim. This means auto insurers are more wary of insuring such individuals.
Today, insurers rely on generalized linear models (GLMs) to mine and examine data with various predictive analytical tools to gain a better insight into their customer base. Before the advent of GLMs, insurers relied on highly aggregated data sets, where rating plans were based on the collective judgment of underwriters and actuaries, while data was simply used as a secondary method to determine where and how to deviate from a common policy.
Modelers now use a variety of methods to create GLMs, with a blend of statistical diagnostics and practical tests applied to a specific insurance company's business principles to predict rates and pricing. The latest enhancement for auto insurance is the use of telematics. Telematics uses an electronic device to gather information about a particular customer's driving habits. The device mines driving data from the vehicle's engine control unit (ECU) to provide insight into a specific customer's driving behavior, which in turn could affect their insurance rate. Usually referred to as scenario testing, the use of this device lets insurance companies use real-time driver information to create policy pricing for a customer. In addition, these devices can reveal specifics about the driver's habits, like whether or not he or she slams on the breaks frequently or commit speeding violations regularly.
Companies like Progressive have been promoting the feature by mailing customers who sign up for the service a small device that plugs into the vehicle's onboard diagnostics port, usually located on the lower left-hand side of the dash board. Then, the device records and sends driving data back to Progressive. The data is collected and interpreted so quickly that customers can actually log in to their accounts and check a running price estimate until a final policy total is produced on the 31st day.
But while tools like these can significantly impact actual car insurance policy pricing, the Insurance Journal points out that predictive modeling tools are now being shifted to provide insurance companies with ways to improve the efficiency of business as well. The article highlights a particular example of how predictive models could help underwriters work more efficiently, and help marketing departments by pinpointing what about their brand attracts new business.
Claims departments also stand to benefit a great deal from the increase prevalence of predictive modeling. In fact, claims departments typically must process more data and information than any other department, so studying the data from claims should help improve a company's understanding of what types of auto claims drive costs higher, what sort of claims should be flagged for potential fraud, and could even lead to being able to weed out potential fraudsters during the underwriting process.