Sanjeev Kumar is the head of the insurance practice at Saama Technologies, a business analytics services company.
Ten percent of the incurred losses and loss adjustment expenses each year in the property & casualty insurance industry are due to insurance fraud, according to an analysis by The Insurance Information Institute (III). Worse yet, the number of fraudulent claims are on the increase—statistics from the National Insurance Crime Bureau (NICB) show a 19 percent increase in questionable claims from 2009 to 2011.
However, most suspicious claims are paid by the insurers; it is estimated that today only one in five fraudulent claims are detected or denied by insurers. Thus insurance fraud costs insurers tens of billions of dollars each year in an industry where margins are thin, and as a result increases premiums for everyone.
P&C insurance fraud may be committed at different points in the transaction, most typically by:
- Applicants when they misrepresent facts on an insurance application.
- Policyholders as they file false or inflated claims (or deliberately perpetrate a crime, such as arson).
- Third-party professionals, such as body shops, that provide services to claimants through excessive billing of vehicle body parts or repair work.
- Employees, such as adjusters, who may be ‘involved’ in the group.
- Agents who may backdate a policy prior to loss date.
Fraud is not just limited to property and casualty insurance. According to the National Health Care Anti-Fraud Association, up to ten percent of the nation’s annual healthcare outlay is lost to fraud and abuse. Fraud in the healthcare insurance industry occurs in multiple forms, such as stolen physician or patient identities, phantom providers and patients, up-coding, unnecessary cosmetic services, false bills, unnecessary diagnostic services, overtreatment, stacked diagnoses and high-fee services.
If insurers can identify and deny fraudulent claims, they not only improve their loss ratio (which increases margins), but more importantly, they also lower future increases in premiums (which gives them a competitive advantage). Fraud analytics addresses this issue by enabling insurers to identify fraud and alert investigators for further analysis.
Fraud analytics solutions use advanced analytics and data science technologies to cull through transactions and flag the ones that have clear and significant risk for further review. The analytics solutions are embedded in the normal workflow to identify potential fraudulent claims as the first notice of loss (FNoL) comes through, without any delay in processing the ‘good’ claims. The flagged claims can then be assigned to the right claims handler, while the rest can be paid quickly. This ability to separate the two is critical because customers have relatively few barriers to switch insurers and an unsatisfactory claim experience in the previous year can easily cause good customers to switch at the time of renewal.
Most insurers today use rules-based systems for identifying fraud. Such systems test and score each claim against a predefined set of business rules and flag the claims that look suspicious due to their aggregate scores. Such business rules may include flagging a claimant who uses several post office boxes, a chiropractor used repeatedly in slip-and-fall incidents, or Social Security numbers associated with more than one name. While such a system is simplistic and easier to implement, these business rules also generate high false positive rates and quickly become obsolete because violators can easily learn and manipulate business rules to their advantage.
However, innovative insurance companies are now beginning to use more advanced techniques, such as predictive modeling, text mining, geographic data mapping and social network analysis for fraud analytics. Statistical-based predictive models analyze characteristics of known fraudulent claims. These characteristics are strong predictors of fraud and are then applied to new incoming claims to forecast the probability that a given claim is fraudulent and flag it. This method can provide adjusters with a quick and easy way to distinguish between suspicious and meritorious claims, with fewer false positives than a simple business rules method.
Predictive analytics, however, are only as good as the data that is fed into them. The quality of results from these predictive models can be improved by bringing together data from multiple sources (see below), including syndicated data, such as credit history and LexisNexis demographics data.
For example, research shows that there is a correlation between credit profile and insurance fraud-motivated arson and similar hazards. One scenario might be where a person with poor credit history changes the collision coverage for an existing auto policy to a lower deductible and then reports a crash within a few days. Predictive models would indicate that the closer the incident is to the date of the policy change, where a deductible is lowered and the credit profile is not good, the higher the potential of a suspicious claim. The accident possibly may have already occurred, and the customer may falsely report the time of occurrence to be a couple of days after the deductible was lowered. Predictive models built using past data are likely to flag this as a suspicious claim.
Text mining capability allows investigators to factor in unstructured data, such as claim adjuster notes, as well as data captured onsite for additional clues of fraudulent behavior. In addition, advanced data mining techniques, such as Social Network Analysis with advanced visualization is being used to provide useful insight into large datasets and identify individuals and service organizations that are participating in fraud.
Fraud analytics technology is not just for identifying suspicious claims, it is also helping underwriters validate information to help eliminate fraud policy application before it occurs. To combat fraud in underwriting, it is important to catch inconsistencies between what the applicant reports and information available from third-party sources.
Fraud analytics allows underwriters to quickly access databases and public records online to verify application details, confirm prior coverage, and discover undisclosed information. Social media, as an example, should be tapped as a big source of input into fraud analytics, because it captures self-declared events, such as a move to a new place or a new driver in the family, which may be different than the information reported when purchasing a policy or filing a FNoL.
With such advanced technologies, today’s insurers can anticipate, detect, and prosecute fraud more quickly and forcefully. And every dollar invested in fighting fraud can return exponential savings to the bottom line.
Insurance companies should take advantage of advanced fraud analytics to anticipate and reduce fraudulent policies and claims in order to improve loss ratios. Reducing fraud through advanced analytics will have a faster and deeper impact on profitability than from improving operational efficiency. There are already some changes at work - innovative insurance companies are beginning to hire less ex-military/police personnel and more statistical modelers in their Special Investigation Units (SIUs).
However, as discussed above, comprehensive fraud detection models incorporate multiple advanced analytics techniques—not just business rules or traditional predictive models. This requires expertise in areas such as text mining, social network analytics and data science. While insurance companies have deep expertise in skills such as actuaries and statisticians, they lack expertise in today’s advanced analytics areas mentioned above.
Also, standard data quality approaches, often used by insurance organizations, need to be modified for fraud detection purposes or else it is possible to wipe out the anomalies that are indicative of fraud. In addition, these advanced analytics models must be refined periodically – usually one or two times per year at minimum—to maintain peak performance, since fraud schemes continuously evolve. While some insurers can build such expertise internally, for others we recommend partnering with analytics specialists. These specialists can help you build and implement state-of-the-art fraud analytics models that are custom to your environment, but can be deployed rapidly.
Evaluate your analytics partners’ ability to achieve fast customization by starting with a framework that contains a pre-built data model that can include inputs from multiple sources such as transactional, syndicated data and unstructured data, as well as a fraud engine with predictive modeling and text mining. It is also important that they have expertise in data science.
Reducing underwriting and claims fraud is the next frontier of significant operational improvement in the insurance industry. It can improve your loss ratios, but more important, provide you a source of competitive advantage by enabling you to compete on premiums in your important segments and win.