Claims fraud is a large-scale problem for the insurance industry, but it is important for insurers to differentiate between “opportunistic” fraud and “organized” fraud. Opportunistic fraud is sporadic fraud activity, often occurring on a one-time basis. Such fraud may involve exaggeration of a claim to make up for losses under a deductible.
In a large-scale survey conducted in 2010, Accenture examined consumers’ attitudes toward insurance fraud and those who commit it. The company found that, among other factors, poor service is a major contributor to opportunistic fraud. The results suggested that more than half of U.S. adults believe that poor service from an insurer is more likely to cause an individual to commit fraud against that company.
The survey also indicated, encouragingly, that consumers themselves still find insurance fraud unacceptable. Only 12 percent of U.S. adults surveyed said they believe it is acceptable to overstate the value of an insurance claim, while five percent approved of submitting claims for items that have not actually been lost or damaged—or for treatments or services not received.
By addressing opportunistic fraud through service improvements, insurers can free up time and resources to deal with the second type of abuse, namely that of organized or institutional fraud. This type of fraud is not linked to dissatisfaction with the quality of service. Sophisticated fraudsters—who are often extremely knowledgeable about insurers’ efforts to detect and deter fraud—represent a prime target for investigation for those insurers seeking to reduce costs associated with dubious claims.
Establishing business rules to identify and screen claims for the most common types of fraud is an important first step in reducing losses stemming from fraudulent claims. More sophisticated tools are available to detect anomalies and determine if and when such anomalies should be referred to special investigative personnel. Some of the rules commonly applied identify policies with more than one or two claims in a single year, or perhaps multiple claims from a single address. Others might pinpoint individuals filing claims against more than one policy, or delays in reporting a claim from the time of the reported accident or injury. Of course, all of these are statistical indicators of possible fraudulent activity.
At the next stage, insurers may employ predictive analytics to assess historical fraudulent claims and identify future indicators of fraud. Predictive analytics combine data, statistical, and quantitative analyses with explanatory and predictive models and fact-based management. This technology can assist insurers in detecting and then referring suspected fraud early in the claim process. Being able to spot suspicious activity early on can thus enable insurers to reduce payouts for bogus or exaggerated claims. Obviously, heading off a fraudulent claim before a payout is preferable to investigating a claim after a payout.
Components of a well-designed analytics suite may include data mining, scoring frameworks, data visualization tools, forecasting models, and statistical analysis. Insight derived from analytics can help insurers make better decisions about taking on certain types of clients to improve overall business outcomes.
Insurers face a difficult challenge in detection: actually knowing when large-scale fraud is taking place. Perpetrators of fraud are always developing new ways to scam insurers. Business rules and predictive analytics can be effective in detecting types of fraud which have taken place in the past, but such techniques may not be able to identify new forms of fraud.
Some of the most interesting work in fraud detection involves pattern recognition, a type of claims analytics that uses algorithms to categorize a range of variables to identify patterns and behaviors that may point to fraudulent activity.
By visually “mapping” the geographic and other connections among objects and people, pattern detection analytics can identify networks that may serve as the basis for additional investigation. For example, a number of claims may involve people living in the same town (sometimes in the same building). The same types of injuries may be treated by the same doctor or group of doctors.
Social network analysis is another rapidly evolving technique for recognizing suspicious activity. Newer technologies include social network analysis, monitoring sites such as Facebook to establish links between possible participants in fraudulent activity. When these connections are visually mapped, they often take the shape of a ring, giving new meaning to the concept of a “fraud ring.”
Intensive mining of data from a variety of sources—including call centers, industry databases, e-mails and letters to the company, adjuster notes, and interviews with individuals pressing a claim—may help establish links and connections between individuals. Network analysis can incorporate Vehicle Identification Numbers (VIN), telephone numbers, and similar data to show behavior patterns meriting further investigation.
In Accenture’s experience, insurers who take an integrated approach to dealing with claims fraud — addressing fraud as it pertains to recovery, loss reserving, workforce training and development, and personal injury management as well as overall customer service — are more likely to be successful in reducing losses associated with fraudulent claims than those insurers who address each problem separately.
A good example can be found in the previously mentioned link between customer service and opportunistic fraud. Insurers who review and improve their claims handling practices, including interface with customers, can help ensure that claims handling contributes to overall customer satisfaction and a reduction in such fraud claims.
At the same time, however, insurers can segment those claims, keeping them separate from the more specialized investigative units and directing these resources toward more detecting more sophisticated, professional practitioners of insurance fraud. Insurers should be able to more fully automate the process of detecting opportunistic fraud.
Successfully establishing claims processes that combine good service with effective fraud prevention rests on three major elements:
First, insurers need good data. Customer background information, such as billing and policy history, history of contacts with the company, and historical claim frequency, should be thorough, accurate, and organized for ready accessibility. Accurate data helps prevent “false positives,” when customers are investigated and found not to have committed fraud. In addition to customer data, claims systems should be linked to external data from sources including CarFax, credit bureaus, provider information, and the ISO. Adding this information to existing client data helps fill out and add needed detail to the picture of the potentially fraudulent claim.
Second, insurers need good people. During the downturn, insurers, like companies in many other industries, cut back on hiring and deferred investments to train their employees. The industry itself faces long-term demographic challenges, as well. The average age of senior claims investigators has been steadily rising, and insurers are having difficulty identifying and training candidates to replace these seasoned professionals. Improved analytics can ameliorate the situation to some extent, by directing claims with a high potential for fraud (and a high potential for return on time invested) to the people most able to handle them.
Over time, however, insurers will need to revisit their overall approach to attracting, retaining, and motivating talent. This can have a tremendous impact at the service level—affecting the propensity of customers to engage in fraudulent activity in the first place—and can also accelerate the process of developing the technologically sophisticated professionals needed for investigative work.
Third, insurers need good technology. The return on investment (ROI) for anti-fraud analytics can be quite high, especially when such investments are made as part of an integrated program to deter and detect fraud at all levels. Analytics, however, are only part of the story for insurers. Mobile technology can assist insurers in capturing data very close to the time of a reported incident — for instance by forwarding a photograph of a damaged car immediately after an accident — as rapid action is a key element to discouraging fraudulent claims. Many insurers are coping with antiquated or legacy claims processing and policy administration systems. Such systems may be deficient in collecting and organizing necessary data, or in providing links to outside databases.
Over and above these problems, however, outdated systems make it difficult for insurers to provide the kind of service that consumers have come to expect from the companies they deal with. This is especially problematic given the nature of personal lines P&C insurance, which is essentially a promise to protect the customer’s most valuable possessions.
While insurance fraud is a pervasive problem, it is worth noting that only 12 percent of the U.S. adults surveyed by Accenture consider that overstating the value of an insurance claim is acceptable. While U.S. consumers may acknowledge customers’ rationales for opportunistic fraud – from tough economic times to poor customer service – the vast majority of customers neither engage in nor condone fraud. And an overwhelming majority of customers (98 percent) feel that it is important for insurers to investigate claims fraud.
Insurers who take the necessary steps and use the appropriate technologies to fight fraud will see virtually immediate returns in terms of lower losses from fraud-related claims. Improvements in customer service will reduce opportunistic fraud, allowing for more intense concentration on organized fraud, with potentially higher returns in claims dollars saved. This may, in turn, help insurers establish a competitive advantage in an always-competitive industry.