Insurance fraud is making headlines as it increases policy prices for consumers, and further squeezes the bottom lines for insurers. Larry Jacobson, insurance expert at FICO, is looking at how technology can be an enabler in making the right decisions to re-gain a competitive edge.
It is virtually impossible to determine exactly how much money is stolen through insurance fraud around the world each year because the majority of it goes undetected. An estimated 15-30 per cent of every premium dollar is lost through premium leakage and fraud putting the price tag for general insurance fraud losses to at least $52 billion worldwide. Insurers often accept fraud as a cost of doing business. However the magnitude of insurance fraud today is startling and urges insurers to act.
Challenging economic conditions have pushed many kinds of fraud to new highs in markets worldwide. In the US, insurance fraud accounts for an estimated 10-20 per cent of insurance premiums; that number climbs as high as 25-30 per cent in other markets, such as Brazil. A single scheme uncovered by the FBI in February 2012 racked up an estimated $279 million in losses. And in the UK, the Association of British Insurers says insurance fraud costs an estimated £2 billion per year. With organised scams involving staged accidents and overstated claims, motor insurance fraud is becoming a particular money-spinner in hard times, and a backlash against rising auto insurance prices.
The fight against insurance fraud is winnable — but are we winning?
Insurance providers are not yet in a position to claim the upper hand. The principal reason for this is the reliance on technology and methodology less advanced than that used in other sectors that experience prevalent fraud, such as credit cards.
Therefore, a comprehensive approach to fighting fraud is critical to prevent mounting losses. The complexity and scope of insurance fraud grows, and the systems used to detect and prevent fraud need to evolve as rapidly as the patterns of misuse they’re trying to identify.
Sophisticated technology can help in this battle. A large number of insurers have already implemented new technologies to find more fraud (64 per cent of participants in a recent survey). However there are still 26 per cent who are taking other approaches, such as adding investigators and investing in new software to speed claims processing overall. A full 10 per cent have not changed their approach to fighting fraud at all over the last several years, which makes even more vulnerable.
Don’t just rely on rules
The most common use to detect fraud is using automated business rules. Rules-based detection is an effective tool for catching known fraudulent activity. But rules, by their very nature, can only be written to identify fraud on known information. Many insurers, for example, have a rule that a claim filed less than 30 days after a policy is written should be flagged and investigated. This seems sensible, but in reality it is precisely the kind of rule that sophisticated criminals will anticipate and outsmart.
This leaves the door wide open for emerging threats that capitalise on the fact that rules-based systems don’t yet know what they’re looking for. Fraudsters are now using more sophisticated methods to test the threshold of tolerance and are gaining sizable gains. So fighting fraud with equal levels of sophistications has become an imperative.
Predictive models are a step ahead
Predictive analytic models on the other hand, use algorithms to look for aberrant patterns in the data that don’t make sense. With a complex, multidimensional analysis these models detect problems that business rules miss, highlighting a systematic pattern of abuse and catching expensive fraud problems long before rules-based claims analysis identifies an issue.
Analytic models continuously learn from the data accumulated from new claims and new emerging fraud trends. As a result they can detect new forms of fraud more frequently as they emerge and before any claims are paid.
Analytic models also give reasons and explanations why a specific claim appears to be suspicious, so analysts can quickly determine whether a claim is legitimate, fraudulent, or worthy of further investigation. Being able to rank claims based on how suspicious they are and to point investigators mo exactly to what is suspicious about them allows insurers to focus their efforts where they are needed most, so they produce the greatest financial impact.
Combating organised fraud with new technology
Combining the two technologies provides insurers with a powerful tool to detect fraud. However, the most advanced technical development helps insurers to add an additional ‘security layer’ top their fraud detection models.
Well-organised criminal rings are penetrating many of today’s most challenging instances of insurance claims fraud, such as the massive increase of whiplash fraud. The UK has been dubbed the 'whiplash capital of Europe' as car insurance costs continue to climb due to fraudulent claims. The Association of British Insurers (ABI) says that one out of every 140 people claim for a whiplash injury each year. These staged accidents can involve large groups of criminals and insurers often struggle to determine genuine claims from fraudulent ones.
Advanced link analysis tools can ease these difficulties by connecting dots that are not obvious. Looking for similarities between claims, involved parties and organisations, link analysis analyses vast amounts of data and finds links between apparently disparate activities. Highlighting connections by as little as a single common item, the technology can uncover previously undetected or unsuspicious instances of fraud. With this methodology insurers can for instance detect whether car crash witnesses testifying in an otherwise unsuspicious claim also provide testimonials to a variety of different claims.
This high level of technical sophistication is essential to uncover organised crime rings that often operate complex scams. The visual link analysis technology helps investigators find patterns quicker.
The analytics-based approach yields dramatic results. By combining predictive analytics, business rules and link analysis, insurers can detect up to 50 per cent more fraud than rules-based systems alone. This is a significant saving considering the $52 billion the industry looses to fraud every year. Insurers see return on their investment within a matter of months. Using this layered fraud detection approach one insurer for instant doubled the amount of fraud dollars detected and found 33 per cent more claims in a fraud ring – just in the first 200 claims that they reviewed.
These are impressive statistics which will certainly deter fraudsters. Insurers are looking at these advanced detection tools but budget and other, conflicting priorities still dominate the reasons insurers cite as barriers to adopting new technologies to fight fraud.
The insurance industry should begin its movement to a more advanced analytical approach as they will find that criminals seek out the companies with the weakest protection. Providers who embrace and understand the intelligent use of analytics will grow market share at the expense of those who don’t.