These are hard times. A lot of bad things happen in hard times.Businesses cut back, and some fail. People worry about losing theirjobs, and some people do. Many people think about where they cancut back on their expenses, and some think about how to get moreincome. Some of these people may decide one way to get more incomeis to file additional claims with their insurance company.

Strictly speaking, fraud occurs whenever a policyholder files aclaim knowingly making false material representations. If Johnsells his car to a cousin in another state and then reports it asstolen, that is fraud. If Mary accepts a role as a whiplash victimin a staged accident and then visits a designated physicaltherapist, that also is fraud. Such actions are easy to label ashard and/or organized fraud–and could be criminally prosecuted withsufficient evidence.

What about Jim saying he needs a couple of weeks more to recoverfrom his workplace back strain? Or Marcia reporting the fire in herliving room destroyed a newer and more expensive sofa than actuallywas lost? These also are examples of fraud but can be characterizedas soft fraud resulting from moral hazard.

The financial incentives for both hard and soft fraud increaseduring tough economic times. Enough people responding to thoseincentives increases the overall level of fraud–in a wayproportionate to the length and severity of the recession.

Every insurer tries to spot potentially fraudulent claims andrefer them to a special investigation unit (SIU). Every SIU tracksthe results of its investigations. And various reports are sent totrade groups and state agencies. However, putting a dollar value oneither the amount of “normal times” fraud or “hard times” fraud issurprisingly difficult.

This is because the aggregate of these referrals andinvestigations is not the same as the total amount of insurancefraud that occurs. There are two kinds of leakage: fraud theadjustment or SIU processes do not discover, and fraud that isdiscovered but not pursued. Undiscovered fraud results from pooradjustment practices or technology. Unpursued fraud results frommore or less deliberate policies that weigh the negativeconsequences of very aggressive antifraud measures against theimpact on claimant satisfaction, compliance with mandated claimspractices, and damage to the insurer's image and marketposition.

So, even after an insurer acknowledges–perhaps to itself,perhaps very quietly–it will not identify and pursue all fraudulentclaims, there still are many things it should do to reducefraudulent payouts. And there are many ways technology can supportthose measures.

Creating automated scores for fraud potential. Any insurerwriting personal lines workers' compensation or health insuranceshould provide its claims adjusters with scores that indicate thelikelihood of fraud.

Updating fraud potential scores whenever new information becomesavailable. As a claim is adjudicated, more information becomesavailable, e.g., police reports or third-party involvement bycertain physicians and attorneys. Scores and profiles should berecalculated to reflect new information.

Selective and intelligent use of external data. Checking thepeople, places, and things involved in a claim against externaldata sources (e.g., known bad actors, vehicles involved in priorclaims, etc.) is a powerful tool for identifying potential fraud.This will involve extra costs, so an insurer must establishguidelines or rules for determining when the cost-benefitrelationship is favorable.

Making indicated referrals to SIUs. This might seem achinglyobvious, but it is not unusual for a relatively small number ofadjusters to account for a large proportion of referrals to theSIU. Conversely, many adjusters make few referrals to an SIU. Whenthis occurs (after making allowance for differences in adjusters'case mixes), this pattern indicates a serious problem. Workflow,rules, and business activity management (BAM) technology canautomate such referrals and enhance claims managers' ability tomonitor actual performance.

Avoiding making too many referrals to SIUs. The number offalse-positive SIU referrals should be minimized. Fewerfalse-positive referrals make the SIU more efficient–and fewerfalse-negative referrals will push the loss ratio down.

Improving the SIU process. The SIU investigation process(receive referral, investigate, take appropriate action) mirrorsthe claims process itself. Many of the same issues apply to both:making correct initial assignments to adjuster and investigators;deciding what external information to obtain or resources todeploy; deciding when and how to complete the process (pay/denyclaim, find/not find fraud, and pursue/not pursue restitution orcriminal complaint).

There are a number of specific technologies that support thesebest practices.

Predictive modeling. This is the use of various statisticalmethods (such as regression, or classification and regressiontrees) to establish the probability a given claim is fraudulent. Atrained analyst or statistician will use several techniques to findwhich model or models provide the greatest level of predictivepower. An important benefit of predictive modeling istransparency–it is easy to identify and describe the factors andthe weights that contribute to the outcome.

Neural networks. This also is a statistical method in whichprogrammed processing agents work in parallel to develop a fraudscore. They have been widely used to determine fraud in the creditcard industry. Neural network analysis can be used with largeamounts of structured or unstructured data (e.g., adjuster notes).Neural networks may do a better job at finding fraudulent claimsthat prior fraud detection methods missed. A disadvantage is it isdifficult to explain how neural nodes work to an adjuster, an SIUinvestigator, or a state Department of Insurance regulator.

Profiling. Although this term has become politically charged inrecent years, it remains a legitimate and powerful technique whenused properly. Profiles are characteristic patterns of behavior andcircumstances. For example, a profile of a valid claim couldinclude a distraught claimant at the time of the first notice ofloss and a more calm and cooperative claimant on subsequentcontacts. A profile of a fraudulent claim could include a calmclaimant at the first notice of loss, followed by severalaggressive contacts in rapid succession. Profiles also are veryimportant for understanding typical and atypical patterns of injury(various injuries resulting from a specific trauma) as well aspatterns of care (two doctor office visits vs. 10 doctor officevisits).

Claims databases. There are several commercially availabledatabases of claims that have been made over many years to manyinsurance companies. One indicator of potential fraud is anassociation of elements in a current claim (such as a claimant, athird party, or a specific vehicle) with earlier claims. Claimsdatabases allow an adjuster (or an automated function within aclaims system) to query these large databases of prior claims todiscover whether there are shared elements with the current claim.One such association or more can indicate a somewhat elevatedpotential for fraud.

Identity matching. A person's identity has multiple dimensions:name, address, phone numbers, Social Security number, occupation,physical appearance, etc. There is no claim-related reason for aperson making a valid claim to change any element of identity.However, people with a pattern of making fraudulent claimsfrequently will change various elements in their identity. The goalof identity-matching technology is to create a complete record ofspecific individuals' interactions with all insurers. A person ofinterest could be a claimant, an injured party, a witness, amedical or rehabilitation provider, or an attorney. For practicalreasons, people almost never change all the elements of theiridentity. Robert Stone may become Rodney Silver, but he will keepthe same cell phone number.

Link analysis. This technology examines the connections amongpeople, places, and things involved in multiple claims. Initialrecognition of recurring connections can be established by theautomated use of search formulas (algorithms). Findings usually aredisplayed graphically, giving an analyst the ability to start witha single element (for example, an attorney) and see what linksexist with physicians, accident locations, witnesses, and vehicles.Link analysis is well suited for discovering or investigatingpatterns of organized fraud.

Other technologies. Business rules or business processmanagement solutions can be used to initiate the calculation orupdating of fraud scores and the subsequent actions adjustersshould take. Additionally, case management tools can be used bymanagers and staff in SIUs. Much like modern claim adjusterdesktops, they can provide a digital repository of all relevantinformation as well as workflow design, control, andmonitoring.

Even in these hard times–or perhaps especially in these hardtimes–insurers are looking at the returns of using technology toreduce fraud.

Donald Light, based in Celent's San Francisco office, is asenior analyst in the firm's insurance group. His research focuseson claims, underwriting and policy administration, reinsurance,product life cycle management, business process and rulessolutions, and accounting. He is the author of reports on policyadministration, claims, underwriting, product development,insurance accounting solutions, and fraud mitigation technology. Hecan be reached at [email protected].