Copyright © 2003 Business Insurance

"Technology Helps In Fighting Fraud" September 2, 2002

by:  SALLY ROBERTS

Advances in technology are going a long way toward detecting the estimated $80 billion in fraudulent insurance claims that are submitted every year.

Today's front-end fraud detection technology not only can assess the fraud risk of individual claims-so called predictive modeling-but it also enables the Insurance Services Office Inc. to provide the industry with a huge database with which to cross-reference potential bogus claims. Identification of such claims can lead to further investigation by insurers or law enforcement.

But while the technology is there, its success relies on individual insurers and their representatives. Although companies are spending more money and resources on fraud detection, including technology, they have yet to make full use of the technology available today, observers say.

"I don't see any company that has nailed the front-end fraud detection piece yet,'' said Patrick Ramsden, vp-insurance solutions for Magnify Inc., a Chicago-based fraud technology vendor. "As people begin to utilize predictive modeling more, it will help to improve the accuracy of fraud detection.''

"The industry is still trying to determine how the technology fits in with their business models and work flow processes,'' said Dan Abbott, vp-information support group for the National Insurance Crime Bureau in Palos Hills, Ill.

Nevertheless, insurers of all sizes consider fraud "a serious problem,'' and more than one-third believe that the amount of fraud has increased over the past three years, according to a survey conducted by the Insurance Research Council and ISO that was released earlier this year.

Experts point out that, although it is widely believed that fraud is on the rise, better detection systems might simply be uncovering more abuse.

"I'm seeing more cases of fraudulent claims, but I don't know if it's an indication of an overall increase or whether it's just being discovered more because of advances in technology,'' said Randall H. Wilson, a partner and national director of fidelity services at RGL Forensic Accountants & Consultants in St. Louis.

Technology advances every year give "insurers a bigger and more powerful database to capture more information to be able to correlate, sort through and track trends, which allows them to see what the trends are and when things don't look right,'' Mr. Wilson said.

Much of today's technological development is replacing the old "rules-based'' system of fraud detection used within the insurance industry, experts say. Under this approach, certain patterns of fraud or schemes were identified, and if a claim followed one of the patterns, a red flag was raised. For example, if a workers compensation claim was filed on a Monday, a red flag might be raised to indicate the possibility that the injury occurred at home over the weekend rather than at work.

The problem, experts say, is that the rules-based system is insufficient in today's more-sophisticated world of insurance fraud.

The rules-based approach looks for only a certain number of the schemes that can be perpetrated, said Steve Biafore, vp-predictive health care group at technology vendor Fair, Isaac & Co. Inc. in San Diego. "There's always another way, and the people using those other ways are not being found,'' he said. "Until you know what those ways are-and they aren't always self-revealing-you can't stop the problem. You need some of these breakthrough technologies.''

"A bad guy, in order to perpetrate fraud, has to make up a story-a story that is going to back up a claim to be paid,'' Mr. Biafore continued. "That story, if it's built well and hidden well, can fly through a payment system and generate a payment that the bad guy is looking to get. That story can never be created with all the data available to the good guys today,'' he said.

"The good guys know everything that everyone is doing. They look at all the injury claims, they look at all the bills submitted by providers, they look at all of the indemnity payments; they have a very complete picture of what's going on,'' Mr. Biafore said. "The problem is, most insurers don't use that complete picture to their full advantage.''

"It's very difficult to access all the available data, because there's a lot of it and it can be very complicated,'' Mr. Biafore said. "And we need to make a decision about who's risky and who's not for fraud.''

This is where predictive modeling technology can help.

Magnify, for example, began focusing on predictive modeling technology a few years ago, Mr. Ramsden said. It's really a mathematical equation, he said, with a historical sample of data run against various fraud data to find patterns. A claim is then ranked on a scale of zero to 1,000, indicating the level of fraud potential, he said.

Mr. Biafore likens predictive modeling technology to a funnel, with data put in at the large end and an accurate assessment of the fraud risk coming out at the smaller end.

In general, predictive modeling software can be used through an application service provider on a company's system, or vendors can take the data from individual companies and run it through their own systems.

Once a fraud assessment is made, high-scoring claims can be set aside for further investigation. These claims also can be run through ISO's ClaimSearch comprehensive database.

ISO's ClaimSearch system was developed three years ago from the merging of industry's leading claims databases for property, casualty and auto insurance, explained Richard Boehning, senior vp of ISO in Jersey City, N.J.

Insurers and others submit claims through their own claims systems or through the ISO ClaimSearch system's personal computer software. The system automatically enters the claims into the database and searches for other claims filed by the same individuals or businesses. The system searches for matches in identifying information fields, such as name, address, Social Security number, vehicle identification number, driver's license number and tax identification number. The system returns reports on any matches to the submitting insurers.

"The more information we get from companies, the better the search and data that we give back,'' Mr. Boehning said.

"We've seen a dramatic increase in the number of claims being sent to us,'' he said. Three and one-half years ago, ISO ClaimSearch had received about 90 million claims; today the database has 290 million claims.

The database will have 400 million to 450 million claims within two years, making it "very powerful,'' Mr. Boehning predicted. "I know it's going to have a dramatic effect.''

© Copyright Business Insurance 2003