The FICO Insurance Fraud Survey makes for disturbingreading for those looking to combat insurance fraud.

While estimates of losses because of fraud ranged from 10% to20%, the majority of respondents predicted a rise in fraud acrossmost categories of personal lines insurance.

Hidden within these figures, but almost impossible to accuratelyevaluate, is the thorny problem of soft fraud. Lost among theheadlines is the plain fact that soft fraud is often committed bynormally honest policyholders. Good policyholders can inflatevalues advertently or inadvertently. A sense of entitlement mayprevail, leading to opportunistic fraud.

Perpetrators are not easy to distinguish from honestpolicyholders. They pay their premiums on time, have good jobs andare unlikely to have criminal records. Unlike hard fraud, a smallbut worrying proportion of the public does not consider soft frauda criminal activity. As many as 68% of respondents in a recentInsurance Consumer Fraud Survey of 2010 feltthat insurance fraud occurred because people believed they wouldnot get caught.

While it is in the interests of both insurers and honestpolicyholders to crack down on soft fraud, it is important thatthis sensitive issue is managed tactfully. It's prohibitivelyexpensive for insurers to fully investigate every single claim,especially when the value is low. Moreover, efforts to detect andprevent soft fraud, which slow down claims processing, may have theunintended consequence of frustrating honest claimants, which inturn can be detrimental to retention levels. Thankfully, frauddetection systems are becoming increasingly sophisticated andinsurers can now apply a range of measures to find a solution.

Detecting soft fraud

In its report, the “State of Insurance Fraud Technology,” Cary,N.C.-based analytics software developer SAS reported that 81% ofrespondents used automated systems employing business rules todetect fraud. Perhaps claims with a value above a certain thresholdare flagged, or claims that are not supported by a police report.Rules-based systems have the benefit of being simple to apply.However, they generate a large number of false positives and maynot be well suited to dealing with soft fraud. Profilingpolicyholders according to criteria such as credit rating or otherpersonal risk factors alone is not going to alert an adjuster toclaims padding.

Anomaly detection — applied by 45% of SAS respondents— can be used effectively for spotting soft fraud. How likelyis it that a policyholder living in an apartment in a high densityresidential area owns a high-powered leaf blower? Is it likely thata 20-year-old driver on an average income would be driving anexpensive sports car? Key performance indicators can be used to setthresholds, which if breached, will trigger an alert.

Comparing the repair or replacement cost of an item to itsaverage retail value can also immediately flag possible claimspadding. For example, a claim for $800 relating to a TV with aretail value of $400 should immediately be flagged for furtherinvestigation. Formal special investigation unit investigations canbe expensive, so be cautious here. Alerts alone help reduce fraudby allowing claims managers to quickly follow up on specific areasof the claim.

Adopting predictive analytics

Predictive analytics solutions were employed by 43% of SASrespondents. With a wealth of information available to insurersabout past claims, it becomes possible to develop a model of futureclaims. What is the average cost of a burglary claim in ageographical area? What is the average value of electronicequipment owned by a family of a certain size in a certain incomebracket? In addition to being more efficient than standalonerules-based systems, predictive analytics can yield fewer falsepositives and can be adapted in real time to incorporate newdata.

The Coalition Against Insurance Fraud has notedthat around half of insurers cite a lack of IT resources as themain stumbling block in implementing anti-fraud technology. As anew generation of anti-fraud solutions are brought to theforefront, more and more insurers will see a positive return oninvestment as they adopt powerful analytical tools to combat softfraud. The potential gains are substantial. Increasing thedetection of soft fraud will not only reduce costs, it may even actas a deterrent for what is, after all, an opportunistic crime.

Joseph Bracken is vice president of product management forNeedham, Mass.-based contents claims management softwarecompany Enservio.