Data Mining: Essential for Exposing Fraud

By K.C. Agrelius and John A.Giknis

Insurers are continually locked in a balancing act between ferreting out fraudulent claims and paying out legitimate claims in a timely manner.

The task becomes particularly difficult during times of economic recession, when insurance fraud tends to spike. As a multifaceted problem, fraud requires a multipronged solution. Technological advances, such as data mining, should be an integral part of any insurer’s detection strategy because they help identify sham claims without encumbering service for regular consumers.

The turmoil of the subprime mortgage crisis, corporate layoffs, and ­fore­closures has many insurers cracking down to avoid being duped by fraudsters. Some reports show property fires have increased as much as 17 percent from the previous year. And as a precautionary measure, some insurers now require their investigative units to review any home fire claim exceeding $25,000.

Understandably, insurers need to take measures to guard against insurance fraud, especially since fraud schemes are increasingly intricate and complex. However, therein lies a dilemma: How do insurers track down fraudulent cases without doubling their staffs or hampering service to regular consumers?

It is neither practical nor prudent for insurers to refer every claim to their investigative units for review. Doing so would result in delayed processing cycles, and this would be a disservice to consumers with meritorious claims that need to be handled promptly. Moreover, insurers just don’t have the resources to increase their staffs because they are facing declining incomes during the current financial downturn.

Detection methods that rely on humans to spot discrepancies and anomalies consistent with insurance fraud are laborious, time-consuming, and expensive. Even the most expert ­inves­tigator is bound to produce inconsistent results because of the sheer volume of claims he or she has to screen. According to the Insurance Research Council, only three of five insurers rate efforts to track down fraud as moderately effective. Some rate their efforts even lower.

The Data-Mining Solution
Business models that incorporate data-mining programs as part of their comprehensive detection strategy possess the ability to uncover previously unnoticed cases of insurance fraud. Data mining adds automation to the fraud-detection process while saving money and increasing efficiency. Claim databases can also be an integral tool for any investigative unit because they provide superior information for investigating fraudulent claims.

In addition to automating fraud-detection strategies, data mining can shift insurers’ fraud-fighting paradigms from a reactive mindset to a proactive mindset. Data-mining programs delve into a vast database and peruse numerous insurance claims to pinpoint repetitions and anomalies consistent with fraud. Insurers can then flag suspicious claims for further review. Indicators of possible fraud implemented in claims databases assist claims adjusters as well as special investigators in the identification of potentially questionable claim activity.

Not only does data mining help uncover fraud cases that might otherwise go unnoticed, it arms law enforcement agencies and prosecutors with enough firepower to arrest, charge, and convict suspected fraudsters. Julia Hearthway, chief deputy attorney general for Pennsylvania, said her decision to prosecute fraud hinges heavily on whether she thinks her office has enough evidence and factual basis to get a conviction. Data mining can uncover the documentation attorneys need to build a solid case. And insurers can accomplish all of this without compromising service for regular policyholders.

Advances in technology have reached a point where claims professionals can quickly see information directly in the electronic files, helping detect and stop insurance fraud without hindering service for legitimate claims. Data-mining programs can also help investigative units detect more incidents of fraud than they could track down using manual methods.

If the current recession worsens, insurers could very well see an increase in the frequency of insurance fraud. That is why it is in the best interest of insurers, policyholders, and the public to continually pursue and implement increasingly sophisticated and effective measures to uncover dishonest individuals seeking ill-deserved rewards.

K.C. Agrelius is general manager of XactAnalysis® for Xactware Solutions (an ISO company), and John A. Giknis is assistant vice president of ISO ClaimSearch® operations for ISO.