The Fight Against Fraud

Insurance fraud is not going away, but the industry may be getting smarter about predicting the future. Many fraud fighting initiatives, including predictive analytics and anti-fraud teams, are having an impact.

“The insurance fraud problem is estimated to exceed US$40 billion globally and is showing no signs of abatement,” says Russ Schreiber, insurance practice leader for FICO, a predictive analytics and decision management technology provider.

Forty-five percent of insurers estimate that insurance fraud costs represent 5 to 10 percent of their claims volume, while 32 percent say the ratio could be as high as 20 percent, according to a new survey of U.S. insurers by FICO and the Property Casualty Insurers Association of America (PCI).

While it has been estimated that insurance fraud accounts for up to 10 percent of property/casualty insurance losses, the survey reveals that some in the industry believe fraud could be more prevalent. More than half (54 percent) of insurers responding to the survey expect to see the cost of fraud in personal lines to increase in 2012; less than 3 percent expect a decline.

The survey also found 67 percent of insurers expect to see an increase in personal property fraud, 65 percent expect an increase in workers’ comp fraud, and 60 percent expect a rise in personal auto fraud. The majority of insurers (61 percent) attributed the increases to sustained economic hardship by policyholders. Only 17 percent of insurers attributed the expected increase in fraud to a rise in the sophistication of criminal gangs. Yet 60 percent expect a rise in workers’ comp fraud rings, and 61 percent expect a rise in auto fraud rings.

When insurers were asked about fraud-fighting initiatives that can have the greatest impact on insurance fraud, predictive analytics was identified as the most effective by 45 percent of respondents.

Predictive modeling is helping to identify possible fraud at the very early stages of a claim, says Steve Armstrong, chief pricing actuary for Chartis Consumer Insurance.

“Claims data is extraordinarily rich with information that predictive modelers could just go nuts over,” Armstrong says. So far, there’s been significant improvement on how to detect fraud early at the point of claims with predictive analytics, he says. “Some people have even started to crack a nut at looking at the potential for fraud on new business by just looking at underwriting characteristics that might lead to a potential fraud down the road.”

Insurers have also found the use of anti-fraud teams for specific books of business (37 percent), link analysis for detecting fraud (31 percent), business rules for stopping known fraud types (29 percent), and external databases (29 percent) as other useful fraud-fighting approaches, the survey said.

Whatever the approach, early detection is critical in the fight against fraud for insurers.