The Detection Triangle

As insurance fraudsters become more sophisticated, the industry must evolve and collaborate to counter increasingly sophisticated fraudulent claims. Fortunately, today’s advanced analytics give organizations the capability to sift rapidly through volumes of data and transactions, allowing the company to decide accurately and in real-time whether or not a claim is high-risk and, if so, what its subsequent course of action should be.

Many organizations are going a step further: they are integrating social network analysis tools into the claims process to help unmask relationships using direct and indirect links existing between seemingly unrelated parties and claims. By using modeling techniques such as predictive analytics, anomaly detection and fuzzy matching, social network analysis enables insurers to visualize customers or transactions in terms of their relationship with other parties and claims.

If you couple these fraud-fighting tools with the advent of improved tools to manage ‘big data’ more efficiently (i.e. the emergence of tools that allow insurers to turn their increasingly vast amounts of information into insight), the concept of ‘more’ has never been better suited for success.

Using these pattern-recognition tools, organizations can now manage enormous pools of data, convert the data into digestible information and then carriers can use this information to make decisions quickly, accurately and effectively. In fact, in Gartner’s 2011 Market Trends Report, big data and pattern-based strategies are highlighted as “the next frontier for insurers.”

Gartner asserts that insurance companies learning to leverage these frontiers in order to fully understand their overall risk profile will gain a competitive advantage.  

New Frontiers in Fraud Detection

Speaking of new frontiers, the insurance industry as a whole is still witnessing an increasing amount and sophistication of waste, abuse and fraud, the latter of which is no doubt fueled in large measure by the financial turmoil gripping the world’s economy. And so, as fraudsters continue to become ever more sophisticated, what’s an insurer to do?

The old adage ‘Birds of a feather flock together’ is certainly true in the context of insurance fraud. A new generation of sophisticated fraudsters is linking to worldwide criminal organizations well versed in the art of flying under the radar. Insurers are therefore relying on more sophisticated tools to help them outsmart the fraudsters.

To this end, the concept of consortium data and the sharing of fraud-related information to tackle fraud are gaining traction in Canada. Insurers are starting to realize that plugging into a rich database of shared, fraud-related information can help detect fraudsters who are making claims across multiple insurance companies.

CGI announced a new fraud detection solution powered by SAS in November 2011. In this solution, cutting-edge technology is combined with rich historical data to help give Canadian property and casualty insurers a more accurate picture of fraud and the crime rings that commit it. Data contributed by participating insurers is combined with an advanced software framework to produce a composite fraud score that can be used fully and consistently across the insurance value chain — from policy inception to claim. The solution is more powerful when insurers integrate the consortium data into their own in-house, fraud-fighting tools. By integrating data from their own various lines of business — i.e. home insurance claims, car claims etc. — into the solution’s consortium data, insurers can greatly augment their ability to detect fraud.

This is welcome news to Ontario’s auto insurance system. The Ontario Office of the Auditor General observed in its 2011 Annual Report that the province’s drivers generally pay much higher premiums than other Canadian drivers do. More specifically, the average injury claim in Ontario of $56,000 is five times greater than the average claim in other provinces, the auditor general says. Insurers have linked a significant portion of their escalating claims costs to auto insurance fraud. Estimates peg the cost of auto insurance fraud in Ontario at between 10% and 15% of total premiums, or in 2010 about $1.3 billion. The bottom line? Fraud costs everyone. By reducing the business costs associated with fraud, insures will be able to pass on the savings to customers.

By better sharing critical information, the Canadian insurance industry will be well positioned to break down barriers and increase awareness across all insurance business lines. Increasingly, the industry will be able to coordinate their efforts to fight fraud and approach the issue in a more systematic way. Imagine the concept of consortium data applied to various business segments, or building a cross-industry collaboration portal? The consequent wealth of fraud data would increase detection capabilities two- or three-fold. We have only just scratched the surface of the possibilities that consortium and collaborative fraud fighting technologies can provide.

The future will undoubtedly see more collaboration and cross-industry support, focused around a rich repository of consortium data that allows for the sharing of fraud knowledge.

The key to success, however, is combining consortium data with in-house data, all enabled by leading-edge analytics technology. Many believe these three areas will provide the fundamental pillars for future fraud detection and prevention. Taken together, they will help paint a clear picture of fraud and prevent and disrupt the activities of fraudsters before they even begin.