At Erie Insurance Group, the foundation for fighting fraud is a business strategy that defines objectives, quantifiable goals, and tactical and operational plans: What is the company trying to accomplish? Does the company have enough resources to react to instances of fraud? How is Erie going to get better referrals to its special investigations unit and become more proactive in identifying fraud?"Erie is keenly aware that our experienced people-our claims adjusters, field investigators, intelligence analysts and information specialist-coupled with our use of fraud detection technology and investigation tools, have made our anti-fraud program so successful," says David Rioux, assistant vice president and manager, corporate security and investigative services for Erie Insurance Group, Erie, Pa.
Rioux, who oversees a group of more than 50 investigators, intelligence analysts and information researchers, believes that technology enables his staff to be more effective and efficient at the work they do. The process starts with a predictive modeling product called FraudFocus developed by Magnify Inc., Chicago. Every day, all Erie claims are scored using a combination of the company's historical data and industry indicators. Suspicious claims are red-flagged and referred to field adjusters. If the claim warrants further investigation, it is sent to fraud investigators.
Erie augments the predictive modeling software with fraud alert notices from the National Insurance Crime Bureau (NICB), a Palos Hills, Ill.-based organization that includes 1,000 insurance companies and law enforcement agencies nationwide. Last year, the group issued 817 reports, called ForeWarn Alerts, that include such information as recent scams the NICB is aware of, status dispositions on NICB cases involving adjudication, requests to insurers to review claims targeting specific types of alleged criminal activity, and law enforcement alerts.
The NICB data is combined with information gathered by Erie's own investigations to create an intelligence watch list. The company's intelligence analysts use data mining software called NetMap for Claims developed by Insurance Services Office Inc. (ISO), Jersey City, N.J., to identify trends, patterns and relationships that are unseen to the eye. The data-mining tool also queries Erie's repository representing five years of claims data to check names, addresses, phone numbers and other identifying information.
Erie also imports ISO ClaimSearch data, a repository of 412 million claims representing 10 years of insurance data, to determine if a person or entity has been flagged for fraudulent activity. Once the company is able to identify its exposure to fraud, the case is referred to a field investigator and the information is sent to the NICB.
Rioux, who worked in law enforcement and had 15 years of experience in field investigation and fraud management before joining Erie in 2001, created the company's case intelligence unit. The first person he hired was a government intelligence officer with experience in terrorism, drug trafficking and military investigations. The second hire was a person who was proficient in using data mining software.
"The analyst component is relatively new to insurance fraud investigations," Rioux says. "The job is to collate information, rule out false-positives and support the field investigators."
The process has worked well for Erie. In one case, Erie received an NICB alert about a medical clinic in New York City that was suspected of billing insurers for durable medical equipment that was assigned to patients but not actually delivered.
Rioux says the company would have had a difficult time quantifying Erie's exposure if it had to manually pull billing records and determine if the clinic had billed the insurer for the equipment.
"Within a few hours of receiving the NICB notice, we queried the system and pulled a good number of claims to determine if we were we billed for the durable medical equipment," he says. "We quickly determined the amount that we paid, identified all of the associated entities such as chiropractors, and packaged the information to our investigator in New York-complete with graphical diagrams and charts."
All of the work that would have taken weeks or months was completed within hours and amounted to more than $200,000 in false claims.
The investment in people and technology has paid off handsomely for Erie. Last year, the company's fraud investigation unit conducted nearly 3,500 investigations, which was 4% higher than a year earlier-and that's despite a 12% drop in claims. In the first six months that the system was operating in 2003, referrals to the SIU rose 28%.
Furthermore, the predictive modeling software flagged 25% of the referrals to the SIU. And those claims reached investigators more quickly than claims that were processed manually. The quality of the referral-a measure indicating an adjuster's ability to articulate the specific reasons why a claim is being referred to the SIU-improved to 95% of claims.
"One of the more interesting aspects of fraud investigations is that your success rate improves the faster a case gets referred to investigators," Rioux notes. Among the 25% of claims that were flagged, the average number of days lapsed from the claim report date was 88 days, and for all other referrals it was 163 days.
Erie had projected a return on its investment of $2 million to $4 million annually in additional claims loss impact over its pre-implementation efforts. (Loss impact is the total claims payments not paid due to alleged claims fraud.) In its first full year of production, Erie realized a loss impact of $5 million, or a 31% improvement over its fraud mitigation results in 2003.
"It's unrealistic to completely eradicate fraud, but you have the potential to make a difference and protect the policyholder," Rioux says. "We take fraud seriously, because it can erode your competitive position. And our moral and ethical beliefs are that you cannot pass the cost on to consumers."
