In law enforcement, the most egregious acts of theft are often solved more quickly than cases of simple theft. That's because law enforcement officials often devote vast resources to bringing high-profile crimes to justice.High-profile insurance fraud also is often solved expeditiously for the same reason-the resources committed to the big-dollar cases are significant. But when it comes to common fraud, insurers have found that the jury's out on their overall ability to identify and curb it.
While these "small" fraud cases appear innocuous, carriers die a "death by a thousand cuts" as a result of them. It's all part of what is estimated to be an $80-billion annual problem in the United States, according to The Coalition Against Insurance Fraud, Washington, D.C.
"Insurers have to be able to capture fraud that occurs under the radar screen," says Lynn Patterson, chief underwriter and senior vice president, at Schaumburg, Ill.-based Zurich Life. "Unfortunately it's a type of fraud that's hard to detect."
One thing that plagues life insurers, Patterson explains, is when individuals engage in policy churning-buying multiple policies at a lower face value per policy. In the context of policy churning, an individual might be compelled to cover up a recent heart condition to ensure policy approval-and at an attractive rate.
Deceitful individuals know that acquiring one life policy for $100,000 from 10 insurers might allow them to circumvent the rigorous application process that a person acquiring a million-dollar policy endures. Obtaining 10 policies at $100,000 each, in effect, fetches $1 million in coverage-all obtained fairly easily and under false pretenses.
When Zurich underwriters activate a solution known as the Insurance Activity Index (IAI), they're able to meld a significant amount of disparate, seemingly unrelated activity within a single search.
IAI, developed by Westwood, Mass.-based MIB Group Inc., provides a cross-industry snapshot of an applicant's recent application activity. Unusual application activity serves as a red-flag indicator of fraud.
"This would lead an underwriter to perform a more intensive investigation about the applicant. They could then decide whether to reject the application or price the product more rigidly," Patterson says.
Data drives effort
Oddly enough, the insurance industry's struggle to reconcile fraud mirrors its efforts to successfully execute another industry initiative-customer relationship management (CRM).
It's not uncommon for one customer to have multiple contracts with one financial services provider. Ineffective CRM is reflected by a provider's inability to identify customer tendencies and to paint a complete picture of that individual. In turn, new selling opportunities are lost. The same applies to insurance fraud: Without the proper tools, identifying fraud patterns can't be cohesively woven together.
Carriers insist that both CRM and fraud detection are two priorities within their operations. Similar to CRM, carriers must invest in fraud technology to support traditional offline methods, observers say.
"Implementing technology to combat fraud is still in its infancy," says Laki Balaji, vice president of property and casualty predictive software solutions for San Rafael, Calif.-based Fair, Isaac & Co., a developer of predictive modeling, decision analysis and intelligence management systems. "Many of the technologies that Fair, Isaac has offered over the years were deployed across vertical markets, such as banking and telecommunications. We've now leveraged these technologies to the insurance space."
The banking industry uses fraud-detection technology to expose fraudulent credit-card transactions. One key to this effort was in the data. "The key to fraud detection is data analysis, which rests on the quality of data and the ability to share data across environments and entities," says Dallas Dawley, assistant vice president, actuarial services data management, Hartford, Conn.-based The Hartford Financial Services Group.
On the claims side, the industry has recently begun to mine data with neural networks for early identification of fraud patterns, Dawley notes, adding that an estimated 10% of insurance industry claims have a component of fraud connected to them.
Losses incurred due to fraud once prompted carriers to pass costs along to existing customers through premium hikes. Many carriers realize that in a competitive marketplace, this is a bad idea.
"On the life side, shrinking profit margins within product lines makes fraud detection and deterrence initiatives critical," says Robert DiAngelo, CIO for MIB Group, which provides information and database management services to the financial services industry. "Ferreting out fraud on the underwriting side enables a company to better price life products."
Consumer applications for life insurance in North America dropped 4.6% in August compared with the same period in 2001, according to the MIB Life Index. Rebuilding this volume is essential. And tools that detect and eliminate fraud are part of the equation.
"We conduct business in a very competitive market that's dependent on new business. Yet at the same time, we face a pressing need to improve our mortality position," says Zurich's Patterson. "Rating new business that is subsequently tainted by dishonesty compromises that position."
Exposing irregularities
Life companies face a different set of circumstances when coping with fraud than property/casualty carriers, which places a different spin on fraud reduction tools.
With property/casualty insurance, for example, an individual who misrepresents information would still be around to deny or defend his or her answers to questions that were asked during the application process, says Fran Marinelli, vice president and chief marketing officer, MIB Group.
Such is not the case with life insurance. That's because the insured definitely will not be around to defend his or her answers to questions on an application or other information that wasn't asked, but which might have had a bearing on accepting or rejecting the risk.
Moreover, after a life policy has been in force for two or three years, in the absence of fraud, the policy becomes incontestable. And the burden of proof is on the carrier to prove fraudulent intent.
This provides further protection to an insured's beneficiaries, especially since life policies may stay in force for many years, and it would be almost impossible to trace back the facts at the time of application, Marinelli says.
Zurich Life has an underwriting unit of about 60 individuals who handle 200,000 life applications per year. When its underwriters use MIB's database to perform a search, the database is directly linked to Zurich's new-business system. A majority of the life companies that are members of MIB operate under an integrated system-to-system data exchange as well.
When it comes to a fraud initiative, members of MIB-many of them direct competitors with each other-understand one thing: Competition might be fierce, but preventing fraud can only be done through cooperation.
The MIB Checking Service enables members to compare an individual applicant's information against information in the MIB database. By way of the exchange, members can detect and deter fraud by applicants who omit or seek to conceal facts that are essential to risk assessment processes.
"We use the MIB checking service that provides the exchange of confidential, coded underwriting information among the members," Patterson says.
But following the rating and issuance of a policy, life companies that detect improprieties often find themselves with less leverage since the policy had already been issued. For instance, why wasn't the impropriety found on the application?
Members activate MIB's Follow-up Service, also known as Plan-F, which provides evidence of nondisclosure on an original application during a two-year contestable period.
"Plan-F provides us a 'second chance' to underwrite. It's really the only source of post-underwriting information," Patterson explains.
Through Plan-F, MIB informs of any new coded reports submitted by other members during the two-year contestable period. These reports can alert to conditions that were undisclosed at the time of the original application. "Had we or another life company known about these conditions, they would have rated or declined the original application," says Patterson.
Under one common fraud scenario, an individual applies for a life policy at Zurich after already obtaining a policy with another insurer. With the Zurich application, the applicant neglected to report a heart condition. However, the applicant did report this condition during an interview with the initial carrier. The MIB database can uncover this history, and Zurich can decide to rate the applicant more stringently or decline coverage.
It's not enough that an applicant is dishonest, but Zurich Life is among many life companies that detects improprieties within its own agent networks by using the database.
There have been instances where independent agents have applied for their own life insurance with numerous carrier affiliates. They "churn" policies-meaning acquire them to draw the first-year commission and let the policies lapse soon after issuance.
Taking liberties
On the life side, the solutions have begun to make a difference on operations reflected in positive return on investment numbers. MIB estimates that it's been able to reduce costs associated with fraud by $1 billion.
Put another way, for every $1 that a life company invests to perform a data check through MIB, mortality costs-on average-are reduced by $46 per application.
Zurich's Patterson says it costs anywhere from $1.75 to $2 per search, which comes out to about $400,000 a year based on new-business volume. "It's a powerful deterrent with a great deal of protective value," he adds.
After years of placing an emphasis on premium fraud both on the life and property/casualty side, carriers have begun to up the ante on tools for claims fraud, including solutions targeting auto, homeowners, workers' compensation and other health/medical-related claims.
Of these lines, workers' compensation has long been considered onerous. Even though an initial claim might be legitimate, individuals often attempt to collect benefits well beyond a rehabilitation time. Using technology that was first developed to address bodily injury claims stemming from auto accidents, Hartford, Conn.-based Travelers Property Casualty Corp. has applied it to address workers' comp fraud, says Ken Jones, vice president of Travelers Investigative Services.
Recognizing the Web as an effective way to foster communication and data exchange efficiencies to eradicate fraud, Travelers developed a Web site in 2000, www.travelersfraud.com.
When a workers' comp claim is filed, there's always a chance that the claim is bogus from the start-that an individual is seizing an opportunity to feign an injury. Exposing a workers' comp claim at the outset had been Travelers strength, thanks to the use of these various tools.
But now the carrier is exposing fraud "downstream"-in the weeks and months following the claim.
"In the past, 90% of workers' comp fraud was detected up front, meaning we detected fraud as the claim was submitted," says Vincent Armentano, vice president of workers' compensation, claims, at Travelers Property Casualty. "Now we're also able to get results in the medical indemnity part of the claim. We regard it as 'soft fraud' where individuals attempt to extend benefits even though they're able to return to work."
The use of neural network technology can enable a company "to take a good amount of industry data and train the networks to identify exceptions to the norm. And we've found that 80% to 90% of exceptions are elements of fraud abuse," explains Balaji of Fair, Isaac.
Fair, Isaac's Claims Advisor for Exception Management is a neural network-driven technology that consists of 50 to 60 variables that interact at any given time, says Balaji. "With all these variables to consider, the technology monitors the claim throughout the process, and if the claim crosses the line, Advisor alerts that the claim is no longer within the norm, but is now an 'exception.'"
Using a combination of data-mining and advanced analytics, the software scores claims from 0 to 1,000. The higher the score, the more likely fraud has occurred.
"If a score is in the 400 to 600 range, it still has not crossed the line," Bajali says. "In this instance, a claimant may receive a letter from the insurer that applies some pressure to return to work. If a score registers around the 800 range, it's a clear indication that fraud is being committed."
Claim scores are posted on an intranet site for claims adjusters to access. After the first week of a workers' comp claim, Claims Advisor continues to drill down into these 50 to 60 variables to update the score. An adjuster can monitor the changing scores and then proceed accordingly.
Online, offline synergy
If there is one thing that insurers have learned about coping with fraud, it's the belief that technology alone can't eliminate the problem, but that the effort must proceed in synergy with human fraud detection competencies.
"We blend technology with the day-to-day activities of our investigators. The technology is built right into their workstations and essentially run in the background," Travelers Armentano says.
"Predictive modeling tools can help pick up what an adjuster might miss from time to time. But the thing about fraud is that there's no silver bullet to purge it. We don't have a situation where we've plugged in this technology solution and reduced fraud 30%," he says.
Cooperation among carriers and other affiliates in the effort is essential. Industry participants say that the good news is that in the year following the 9/11 terrorist attacks, "we've seen a great deal of stronger and expanded cooperation. We used to have pockets of good cooperation, but it's vastly better," says Armetano.
