Fighting Fraud the Smart Way

At first, it looks like any other auto accident at a busy intersection, with the total insurance payout amounting to thousands of dollars. Later, both cars are found to be part of a carefully staged collision orchestrated by players known to have engaged in insurance fraud.

But recovering the money will not be easy. In the war against fraud, the criminals are winning.

Fraudulent acts against insurance companies are on the rise, and most go undetected. Fraud accounts for as much as 10% of losses incurred by insurance companies. It takes many forms. For example, an organized crime ring may take advantage of no-fault clauses in policies, or a distressed property owner may see no other way to pay off his or her loan.

For too long, the insurance industry has accepted fraud as a cost of doing business. Historically, pinpointing claim fraud required a significant amount of staff hours that could not be justified based on the returns. However, with combined ratios climbing well above 100%, operational costs already cut to the bone and 0% interest rates all but eliminating investment income, insurance carriers can no longer opt for acceptance.

Fortunately advances in technology and affordability, coupled with skyrocketing combined ratios, have changed the cost equation. Now, finding fraud — and more important, preventing it — is not only feasible, but necessary.

Carriers can now implement technology previously available only to deep-pocketed, crime-fighting organizations such as the CIA, FBI, National Security Agency and RCMP. Indeed, more than 35% of property and casualty auto insurers name fraud detection as a top investment priority in 2012.

Change is already underway. In its April 2011 budget, the Ontario government signaled its intention to arm insurers with the tools to combat fraud and excess billing. By adopting a proactive analytics approach, Manitoba Public Insurance realized almost $10 million in total fraud savings last year.

TARGETING FRAUD OVER THE LIFECYCLE

Fraud prevention is especially effective at two key points: underwriting and claims intake. Fraud can be prevented if the identities, entities or behaviours of the perpetrators buying the policy can be identified. Entity analytics and predictive analytics make this possible.

Underwriting

If the named insured, vehicle or lien holder has been involved in previous claims, technology can identify which policies are being purchased for the purpose of conducting a staged accident. In real time, the script used by underwriters and intake specialists can be changed to let fraudsters know they are exposed. It may not be enough to prosecute, but might drive them to other carriers with less sophisticated means of detecting fraud.

Claims Intake

The same technology can confirm a person’s identity, the relationships between claimants, and the activities and names of individuals and businesses involved in the claim. Armed with insight about the individuals and intent, one could change the questions asked of risks or claimants to determine if the claim is fraudulent. If the claimant knows you suspect him or her, this may be enough to discourage the party from proceeding. If it is a legitimate risk or claim, the party will consider these questions part of the normal process.

Fraudsters are very good at masking their identities from underwriters. However, technology can examine hundreds of elements across millions of records to provide real-time alerts. Similarly, predictive analytics can look at hundreds of rules and conditions in real time that might indicate fraud.

To take a fictitious example, Bill Smith reports a claim. B.J. Smith was a witness in a separate, similar claim last month. John Smith was a passenger in a claim three months ago. Each has similar, but slightly different addresses and phone numbers, but the same date of birth. By examining these entities, it can be reasonably determined ‘the Smiths’ are in fact the same individual. The intake specialist might miss this information, but entity analytics will not.

An alert is fed into a predictive engine along with other data to determine if sufficient evidence exists to challenge the person reporting the claim with a few more details. The intake specialist can determine whether this is the same ‘Bill Smith’ involved in a previous claim, or ask him if he would like to update to a single address. Organized fraudsters are very sensitive to unusual responses. They would likely drop the claim and go to easier marks rather than risk that they have been identified.

These two activities alone can dramatically reduce fraud at the point of submission, before it even gets to the adjuster. Once these connections become known, the carrier might be removed from an organized fraud ring’s hit list.

ACROSS THE CLAIM EXPERIENCE

A smarter, more holistic claim fraud program will not stop at the above. If the claim does reach the adjuster, tools can be attached to existing claims administration systems that assist in analyzing a wide array of data sources — both internal and external, including social media and public data — to determine if anything is out of the ordinary. These can be run in the background, thereby reducing the adjuster’s workload and providing alerts if anomalous activity is detected.

Carriers should regularly churn their data to search for patterns and associations in the data. As a carrier becomes more equipped at finding fraud, it will develop a richer set of rules and known networks to identify fraud that has already been reported. This helps find additional fraud, and provides a better set of rules for searches to prevent fraud.

As fraud is identified, modern visualization and investigation technology can accelerate the time it takes investigators to piece together the elements of a fraud ring. This can often happen during an afternoon, as opposed to requiring 18 months of painstaking analysis.

The final element of a smarter claims fraud program is in the reporting, monitoring and visualization of claim fraud data. A powerful tool is geo-spatial mapping. Here, the ratio of bodily injury insurance claims to non-bodily injury claims is established to identify where a fraud ring might be operating (claims fraud involves more soft-tissue injury claims than a carrier would see under a normal baseline). Once identified, the claims can be examined to find a network of interconnected entities. Carriers can then establish the behaviour rules and individuals involved, alerting them to situations in which these individuals buy new policies or report new claims.

CONCLUSION

In today’s challenged market, carriers can no longer accept that 10% of their incurred losses leaked via fraud are just a cost of doing business. The primary defence against fraud should no longer be limited to educating the adjusters on what to look for and staffing a special investigation unit. These activities are still needed, but they are not sufficient to reduce the significant cost of fraud.

The time has come to leverage available and affordable technology to implement a comprehensive, smarter claims fraud program. Such a program would not only identify fraud, it would predict, and ultimately prevent, fraud from occurring in the first place.

Having sufficient data to prosecute is not always required to prevent fraud. Sometimes just letting the perpetrators know that you are aware of them will cause them to shift their attention to lower-hanging fruit, including those carriers that do not have the same level of sophistication.