Insurers are always on the lookout for ways to spot fraud, and with good reason. The instances of fraud, and the variety of fraudulent activities, continue to rise.
According to the National Insurance Crime Bureau's (NICB) ForeCAST Report, the number of questionable claims increased to 48,887 in the first half of 2011 from 46,766 in the first half of 2010 and 41,309 in the first half of 2009. That represents an increase of 18.3% over the two-year period.
To help snuff out fraud, insurance claims departments and their special investigative units (SIU) are increasingly leveraging analytic technologies to enable them to more quickly, easily and effectively identify potential cases of fraud.
"Modern claims solutions often include scoring mechanisms. Combined with workflow and task generation, insurers can use business rules with scoring mechanisms to identify 'red flags' and automatically refer claims to special investigative units," says Karlyn Carnahan, principal at consulting firm Novarica. "This provides a convenient way for patterns to be identified and files to be referred," she says.
Firms can look at data across business units to identify potential fraud, and examine both claims and underwriting data across lines of business to highlight suspicious claims, adds Carnahan. For example, they could identify a business interruption claim reporting wage loss for more employees than stated in the workers' compensation underwriting file.
"While some of this is done manually, increasingly technology is being used to identify links using sophisticated analysis tools," Carnahan says.
MetLife Auto & Home, a subsidiary of MetLife Inc., has been aggressively going after claims fraud via technology such as analytics, says John Sargent, director of MetLife's SIU.
Back in the days when the company relied solely on information from claims representatives based on their observations and conversations with claimants, MetLife missed a lot of opportunities to identify and stop fraud, Sargent says.
But with the use of applications such as predictive modeling, search engines and automated business rules to analyze claims and patterns, the company has become much more successful in its fight against fraud. Thanks to the use of technology, MetLife Auto & Home has nearly tripled the number of fraud investigations it conducts per year, without the need to increase staff, Sargent says.
One of the key technologies in MetLife's anti-fraud arsenal is an application called Fraud Evaluator, which the company co-developed with Computer Sciences Corp. to identify medical provider, attorney and repair shop fraud.
MetLife uses the system, which incorporates predictive modeling and identity search capabilities, to analyze data throughout the life of the claim. Within the first six months of deployment the software accounted for more than 20% of the referrals investigated by the special investigative unit and the company had realized a significant return on its investment.
Fraud Evaluator discovered claims involving medical providers, attorneys and repair shops that were suspected in multiple incidents of fraud, and it identified one individual with four active claims, three of which were for auto theft.
With the additional information gleaned from MetLife anti-fraud applications, "we can now better differentiate suspicious claims from the truly legitimate ones so we're not wasting our time," Sargent says. "It's much quicker than in the past because we can concentrate our efforts on looking at the suspicious claims."
In its latest efforts to battle fraud via technology, the company is making greater use of text mining of claims data to enhance models and look for patterns that might indicate fraud. It's also using information such as weather data to help spot fraud.
For example, after a recent hailstorm in Arizona that resulted in a large volume of claims, the company used weather data to create a color-coded map that showed whether or not claims came from areas that were likely hit by the storm. Those that came from areas well outside the storm's range were assigned to investigators for potential fraud.
A growing number of insurance companies are investing in these predictive analytics tools, says Russell Schreiber, VP, Global Insurance and Healthcare at FICO, a provider of predictive analytics products. "We are observing a dramatic increase in appetite for this technology among insurers," Schreiber says. "The focus we're seeing in the marketplace has shifted from spending time on the evaluation of the technology to acquisition and implementation," he says.
Nationwide Insurance built a point-of-sale, multi-factor consumer data verification process for auto and property to combat fraud. "By verifying consumer data with public information, we have been able to identify inconsistencies and mitigate fraud attempts early on in the insurance life cycle," says Lynne Brady, associate VP, Nationwide Special Investigations.
The company was an early adopter of business intelligence/analytics, Brady says. "We were one of the first to deploy a sophisticated predictive analytics tool on our claims data, helping to identify potential fraud elements and assist our claims partners in properly adjusting the claim," she says.
Nationwide is examining more sophisticated technologies that bring all of its analytics tools together in order to perform link and networking analytics, Brady says. "Through link analysis and data mining we can bring together people, vehicle, claims, underwriting, billing, third-party and medical provider data and see how they interact with one another," she says. "This view will allow us to better identify fraud rings and group fraud activities."
Analytics has also helped the special investigations unit at property/casualty insurer CNA in Chicago with its anti-fraud activities. The unit recently began using the Fraud Framework predictive analytics tool from SAS, which runs open claims data through predictive models in the workers' comp, general liability, commercial auto and commercial property lines of business and scores files for fraud potential.
The claims are then reviewed by the special investigations unit and either accepted for investigation or rejected, says Tim Wolfe, director of the SIU, claim administration at CNA. The tool also identifies patterns of fraud and social networks within those claim files, Wolfe says. It has the ability to flag entities already under investigation on existing files.
The social networking component of the SAS Fraud Framework "will provide a 'link analysis' type approach to detecting claim fraud, but it is not scheduled for implementation until later this year," Wolfe says.
Since 2007, CNA has been using a link analysis tool called i2 Analyst Notebook from i2 Group. The product enables the user to pull data from disparate sources and analyze it to detect patterns of fraud across multiple claim files, often by medical providers and organized rings.
"Since implementing the i2 Analyst Notebook and hiring a team of full-time intelligence analysts to operate the tool, we can identify connections between claims and providers on a large scale. We can now run both reactive and proactive analytical studies, identify patterns of fraud and the modus operandi of the perpetrators much faster than using conventional methods to mitigate claims, and implement safeguards to close loopholes in our operation," Wolfe says.
In one recent case, through the help of discoveries made using i2 and other methods, the unit was able to reduce $3.8 million in suspect provider billing to a mere $35,000, Wolfe says. It's those kinds of results that will keep insurers on the lookout for the latest anti-fraud technologies, experts say.
Fortunately for insurers, emerging technologies are delivering more ways to identify potential fraud. New sources of data are becoming available to insurers, says Carnahan. "For example, video and audio search capabilities or face recognition technology is being used to better [leverage] what used to be unstructured data in a more consistent manner," she says, adding that another source is data from networked sensors. "Whether it's reviewing driver behavior at the time of an event or looking at sensors on buildings, bridges or even data available from mobile devices, carriers have a growing and abundant source of data emerging."
Bob Violino is a business editor and writer based in New York.
