A successful fraud unit needs experts on claims handling, according to John Elder, chief scientist with Elder Research. “You need someone who knows what a normal claim looks like to work closely with someone who knows mathematics,” he says. “You don't see a lot of success if you give a tool to someone who doesn't know insurance. You need to have those parties working together. A small pilot project usually is a good idea. You can show tangible results and quantify what the potential return on investment is.”

MetLife Auto & Home began data mining to fight fraud about five years ago, reports Sargent. Analysts defined the searches and analyzed the results. They then made the determination whether the claim was suspicious enough to launch an investigation. “We had great results with that, but we felt we could automate,” he says.

The carrier co-developed the tool, known as Fraud Evaluator, with Computer Sciences Corporation. “We were able to be part of the development process so we got everything we wanted,” explains Sargent. The system analyzes the carrier's claims every night. One of the analysis portions of the product includes a model that compares current claims against a claim that previously had been investigated and was found to be fraudulent, he says. Over the last five years, MetLife Auto & Home continued to enhance the product with upgrades and a new data-mining tool within the product.

Different Strokes

Every company does fraud detection differently, Sargent points out. The course MetLife Auto & Home chose was to have the claims scores reported to the SIU where one of the unit's analysts would determine whether the case warranted an investigation. “If you believe the industry numbers that up to 10 percent of claims are fraudulent, then 90 percent of claims are legitimate,” Sargent notes. “What we want our claims reps to do is concentrate on providing good customer service and expert claims-file handling. To burden them with one more additional task, added to the long list of things they already have to do, didn't seem to make sense to us.”

The claims adjusters are not completely off the hook. Sargent acknowledges the adjusters still have the responsibility to detect fraud, and the company remains aggressive in providing the claims staff with training in how to do that. “We felt if we brought the right people into our analyst unit–people with significant claims experience–they would look for what we wanted them to look for and the claims reps could work on 90 percent of the cases that are legitimate,” he says.

While Erie is going from a reactive to a proactive state thanks to technology, Rioux adds the carrier still will continue to be reactive. “The best line of defense for fraud always will be the front-line claims handlers,” he states. “No amount of technology or intelligence information is going to replace what the best fraud indicators do for the industry. Most of the time, [those indicators] are the gut feelings and instincts of the claims handlers. We augment that significantly by getting [claims handlers] tools to steer them to certain claims that have a greater propensity for fraud.”

Predictive Modeling

The whole concept of data mining within the insurance industry is gaining more attention, not only in the area of fraud but also in the area of predictive modeling, according to Steven Sumner, director in the insurance advisory practice with the consulting firm PricewaterhouseCoopers. “Insurance companies are beginning to realize they need to move away from the traditional way of doing business and find other solutions to their problems, especially in the area of fraud because it costs the industry billions of dollars,” he says.

Many business tools give carriers new and better insights into what has happened in the past, Elder asserts. “It's just better than a guess,” he says. “It's valuable for where you have to make decisions. Hopefully, the things you discover in the analysis of the data help you modify your procedures so you maybe add a certain question or two when a particular issue comes up.”

For Erie, Rioux indicates the carrier was one of the first to take on both modeling and analytics. “We wanted to do some predictive modeling and make that a tool for the claims handler to better detect potential fraud,” he says. “At the same time, we needed to work smarter on the back end through data analytics.”

First, Erie wanted the ability to check cases of fraud so they don't slip through the cracks. “The goals we set were quite simple,” says Rioux. “If we were going to do fraud technology and move our SIU operation to the next level, we wanted to detect fraud earlier in the claim life cycle and as soon as possible. The longer time goes by, the harder it is to prove something. The older a claim is and the more people have established their stories, the harder it is to prove [fraud]. If you can catch the [potential fraud] earlier in the claim life cycle, you have greater success.”

Erie's second goal was to get better quality referrals–claims that are suspicious–in the hands of the SIU as soon as possible. The third goal was to determine whether there was more fraud being committed, and if so, the carrier wanted to find it.

Even though not all companies employ predictive analysis, Rioux notes the science has been used for a long time in insurance, particularly among life insurers. “Actuaries can tell you how many people are going to die and where they are going to die; they just can't tell us whether it's you or me,” he says.

Erie has found since it began implementing predictive analysis, it has experienced a good return on investment, according to Rioux, who credits the technology in helping an adjuster to identify claims that need to be investigated. “It's not an exact science, but for every 15 notified claims that have scored high, we get one good quality, questionable claim that needs to be investigated,” he says. “Without technology, it would be more in the area of one in 200 to 300 claims.”

Rioux reports many companies he's been in communication with are in the process of researching predictive modeling and other fraud technologies, and many others currently are deploying systems. “It's growing quite rapidly,” he says of the field. “At one point, the industry as a whole questioned whether predictive modeling would work, but I think people are starting to see there are clear results. Each company has to look at ways to make predictive modeling even better.”

Erie annually rebuilds its models, and as the models take in the previous year's results, they become more intuitive. “This is our second rebuild cycle, and the models we've just deployed for 2006 are better than the prior year's models because we have more known outcomes,” Rioux says. “The more known outcomes we have, we can use them to build a more intelligent model.”

Sitting It Out

There still are insurers that don't participate in the fight against fraud, but Sumner believes those holdouts are beginning to see the price tag associated with fraud. “The number I read recently was $40 billion just for the property/ casualty industry,” he says. Sumner also contends regulators are coming down harder on the insurance industry to do something about fraud. To wage the battle, insurers are looking at their technology and their data. “[Insurers] are getting more sophisticated at it,” Sumner observes. “They are generating reports that help identify the areas where they can take a closer look at some of the claims-handling processes, and they also are doing more on the underwriting side.”

One example of what many carriers can do is to automate claims files to capture the data as it is being reported. “Unless you have the data in [the system], it is very difficult to do your data mining and extract the data you need to investigate fraud,” says Sumner. The insurance industry is going to spend somewhere around $50 billion on antifraud technology by 2010, he estimates, pointing out that's more than two or three times what was spent at the end of 2005.

Helping Hand

“Today, fraud has become more complex,” comments Rioux. “It's difficult to investigate totally from field investigators alone without the support and the backing of some information intelligence.” Erie is a member of the National Insurance Crime Bureau (NICB) and the Coalition Against Insurance Fraud. The NICB works with carriers to investigate fraud and develop intelligence. From that intelligence, information generated provides alerts for carriers. In the past, Rioux felt some of the intelligence reports were too vague and lacked what he calls “actionable intelligence.” Today, though, the reports are more specific. The NICB alerts along with other information are part of the intelligence watch list Erie has created. “We put all that intelligence in electronically, and it constantly screens our system moving forward,” he says. “Sometimes we find fraud activity is not with us today, but it may be with us next week, next month, or next year.”

The NICB has employed a strategic and tactical information unit for about five years, according to Aaron Soline, manager of the strategic analysis unit for the NICB. The unit's goal on the strategic side is to supply analysis on new trends and forecast modeling for the industry. The tactical analysis side of the unit provides direct case support.

The NICB has multiple data sources it uses and a number of different programs depending on the data source. The major data source is the ISO claims database.

In the strategic analysis unit there are a number of predefined questions that everyone wants answered. “One of the big things we deal with is auto theft,” says Soline. “People want to know where stolen vehicles are traveling to, when they most likely are to be stolen, and where they are most likely to be recovered.” Those are relatively easy questions, he points out. “You take your data set, and you insert one of your queries, and you usually can find the answer to it,” says Soline.

The second side is more proactive with analysts looking for abnormalities in the data. “You take a large data set, and you look for the stuff that not everybody else is doing–the one vehicle that is traveling the other way while all the other stolen vehicles are going to a different area,” Soline says. “You try to identify the trend before it becomes a trend.”

Share and Share Alike

Elder is a proponent of sharing as much data as possible. “A precious commodity is a known and labeled case of fraud,” he says. Whenever a company mines data, he adds, it needs to have examples of fraudulent activities in order to find more like them and to build patterns that help predict where the next case is going to be. “Data mining needs to bootstrap up from some examples of existing fraud,” he says. “On the one hand you don't want fraud, but if you've got it, you want as much of it as possible labeled so you can identify and squelch more of it.”

One of the key components to a successful data-mining project is whether the gain is able to be leveraged, Elder claims. With insurance fraud, he maintains that clearly is the case. “If you can stop fraud, the dollars [saved] come straight to your bottom line, and that quells a potentially exponential loss that occurs if you don't stop it,” he says. Elder also suggests companies have to have someone high in the organization who believes fighting fraud should be a priority.

The Future

From Sargent's perspective, his company has been very aggressive when it comes to fighting fraud. “We've invested significant capital and personnel resources to developing and implementing the systems, providing the staff to make sure [the systems] are run properly, and ensuring there are enough investigators to do the investigations,” he says. From an industry perspective, though, Sargent rates the work being done as worthy of a grade of C-plus or a B. To improve that grade requires companies to work closer than they have in the past. “That's the next hurdle we have to overcome,” he says. “The industry as a whole doesn't seem to come together as well as it should.”

More universal access to data is where the future lies in fighting insurance fraud, Soline indicates. “A lot of companies have individual databases they look at, and a lot of people are doing their own analysis,” he says. “Once you start combining those different data sets and people start looking at them as an aggregate as opposed to just their individual concern, I think they'll start finding a lot more [fraud].”

As an example, Soline mentions a medical clinic might be committing just enough fraud against one insurance company to stay under the carrier's radar, but if investigators had the ability to study the data from multiple companies, they might find a pattern. Most companies enter their data into the ISO claims database, and Soline says he is glad to see ISO is going to a universal format that will provide richer data. “As more companies convert over to [universal format], I think [data] will become more valuable,” he predicts. “But I'm not talking just about claims data. There is a ton of information that's available. The more [carriers] share, the more productive everybody is going to be.”

Improvement will come with the development of electronic claims files and having the adjusters use them, Sumner believes. “As the industry starts to figure out how to overlay its underwriting with its claims to start drawing parallels, that's when you are going to start seeing things evolve,” he says. “You are going to see the industry using its claims data to help in predicting underwriting results and the whole decision-making process. As the industry invests in technology, it's not only for fraud, it's for other cost-efficient measures.”

As for the industrywide data, Sargent contends the data that resides within ISO is extremely valuable, but he feels allowing the industry to capture additional data as it pertains to suspicious claims would increase the success of SIUs significantly. “If companies can share not only the fact there was a claim but the fact there was a fraud investigation by the SIU, that's going to add some significant value to the industry as we move forward,” he says. “We think there's real opportunity with the data that sits out there that's been untapped.”

However, there is a resistance in most companies in sharing data, Elder believes. “People see only danger in sharing their data,” he says. “The guardians of data tend to be extremely protective of it even with peers within their own organization.”

@Global, Inc.

Breckenridge, Colo.

303-377-2608

www.atglobal.com

Accenture

Murray Hill, N.J.

267-216-1049

www.accenture.com

Adjustable Software

Markham, Ont., Canada

905-415-0664

www.adjustablesoftware.com

ADP Claims Services Group

Elk Grove Village, Ill.

847-718-3381

www.adpclaims.com

AGO Insurance Software, Inc.

Mt. Arlington, N.J.

973-770-3200

www.agois.com

Allegient Systems

Wilton, Conn.

203-761-1289

www.allegientsystems.com

Allegis Communications, Inc.

Seattle, Wash.

800-566-6112

www.allegisteam.com

American Management Systems

Fairfax, Va.

800-255-8888

www.ams.com

Apex Data Systems.com

Tucson, Ariz.

520-298-1991

www.apexdatasystems.com

Attus Technologies, Inc.

Charlotte, N.C.

704-752-6249

www.attustech.com

The Bluebook International

Lake Forest, Calif.

888-425-8326

www.bluebook.net

Bramerhill Technology Group

Norcross, Ga.

770-368-3255

www.bramerhill.com

Bridium

Atlanta, Ga.

920-451-9777

www.bridium.com

Brightwork, Inc.

Chicago, Ill.

773-529-1100

www.brightworkinc.com

Burkitt Computer Corp.

North York, Ont., Canada

416-642-6159

www.bccorp.com

Business Software Solutions

Mt. Laurel, N.J.

856-866-5601

www.bssnj.com

Castek

Toronto, Ont., Canada

416-777-2550

www.castek.com

CCC Information Services

Chicago, Ill.

800-621-8070

www.cccis.com

CGI

Andover, Mass.

978-946-3343

www.cgi.com

ChoicePoint

Alpharetta, Ga.

800-342-5339

www.choicepoint.net

ClaimIQ

San Francisco, Calif.

415-498-2415

www.claimiq.com

Clear Technology, Inc.

Westminster, Colo.

303-583-4150

www.clear-technology.com

Computer Sciences Corporation

Financial Services Group

Austin, Tex.

800-345-7672

www.csc.com

Corporate Systems

Amarillo, Tex.

800-858-4351

www.csedge.com

Covansys

Farmington Hills, Mich.

913-469-8700

www.covansys.com

Cunningham Lindsey

Lewisville, Tex.

214-488-5139

www.cunninghamlindsey.com

Decision Research Corp.

Honolulu, Hawaii

808-949-8316

www.decisionresearch.com

Delphi Technology

New Brunswick, N.J.

732-418-0008

www.delphi-tech.com

Docucorp International

Dallas, Tex.

800-735-6620

www.docucorp.com

DWL

Toronto, Ont., Canada

416-364-2045

www.dwl.com

EasyLink Services Corp.

Piscataway, N.J.

732-652-3589

www.easylink.com

Ebix, Inc.

Atlanta, Ga.

678-201-2023

www.ebix.com

E-Claim.com

Gretna, La.

504-433-9599

www.e-claim.com

EigenSoft

Portsmouth, N.H.

603-430-8032

www.eigensoft.com

Fair Isaac & Co.

Irvine, Calif.

949-655-3300

www.fairisaac.com

FARA

Mandeville, La.

800-259-8388

www.fara.com

FileNet

Costa Mesa, Calif.

714-327-3400

www.filenet.com

FINEOS

Cambridge, Mass.

877-893-7904

www.fineos.com

First Notice Systems

Boston, Mass.

800-310-4367

www.firstnotice.com

Fiserv Insurance Solutions

Cedar Rapids, Iowa

800-943-2851

www.fiservinsurance.com

FLIR Systems

North Billerica, Mass.

978-901-8105

www.flirthermography.com

Fuego, Inc.

Plano, Tex.

972-801-4218

www.fuego.com

GAB Robins

Parsippany, N.J.

973-993-3400

www.gabrobinsna.com

Genelco Software Solutions

St. Louis, Mo.

800-983-8114

www.genelco.com

GenSource Corporation

Valencia, Calif.

661-294-1300

www.gensourcecorp.com

Group 1 Software

Lanham, Md.

301-918-6381

www.g1.com

Guidewire Software, Inc.

San Mateo, Calif.

650-357-9100

www.guidewire.com

Hawkins Research

Murray, Ky.

800-736-1246

www.powerclaim.com

Hyland Software, Inc.

Westlake, Ohio

919-462-8505

www.onbase.com

ILOG, Inc.

Mountain View, Calif.

650-567-8000

www.ilog.com

Infinity Systems Consulting, Inc.

New York, N.Y.

212-541-7602

www.infinity-consulting.com

Infoglide Software

Austin, Tex.

214-357-4636

www.infoglide.com

Information Builders

New York, N.Y.

212-736-4433

www.informationbuilders.com

Injury Sciences

San Antonio, Tex.

877-979-7378

www.injurysciences.com

The Innovation Group

Danbury, Conn.

203-743-6000

www.tigplc.com

Insurance Data Processing

Wyncote, Pa.

800-523-6745

www.idpnet.com

Insurance Services Office

Jersey City, N.J.

800-888-4476

www.iso.com

InsureWorx, Inc.

Emeryville, Calif.

510-596-1300

www.insureworx.com

InsurSys

San Francisco, Calif.

415-975-0966

www.insursys.com

ISCS

San Jose, Calif.

888-901-4727

www.iscs.com

JCRS Inland Marine Solutions

Oakland, Calif.

510-444-4811

www.jcrs.com

Jury Verdict Research

Horsham, Pa.

215-784-0860

www.juryverdictresearch.com

JW Software, Inc.

St. Louis, Mo.

800-828-6120

www.jwsoftware.com

Magnify

Chicago, Ill.

312-214-1420

www.magnify.com

MapInfo

Troy, N.Y.

800-327-8627

www.mapinfo.com

Marshall & Swift/Boeckh

New Berlin, Wis.

800-285-1288

www.msbinfo.com

Metropolitan Reporting Bureau

Philadelphia, Pa.

800-245-6686

www.metroreporting.com

MFX

Toronto, Ont., Canada

416-385-4800

www.mfxfairfax.com

Mitchell International

San Diego, Calif.

800-854-7030

www.mitchell.com

MountainView Software

Layton, Utah

888-533-1122

www.claimzone.com

n'site solutions

Urbandale, Iowa

888-282-6596

www.nsitesolutions.net

OpenFlex

Los Angeles, Calif.

213-252-2360

www.openflex.com

P&C Insurance Systems, Inc.

New York, N.Y.

646-442-8810

www.pcisvision.com

PayFlight

Clinton, Ill.

309-834-1590

www.payflight.com

Peak Performance Solutions, Inc.

Hoboken, N.J.

410-531-8528

www.peakpsi.com

PCMS

Dallas, Tex.

972-855-3515

www.pcms.info

ProcessClaims

Manhattan Beach, Calif.

310-937-4040

www.processclaims.com

Quest Claims Services

St. Johns, Mich.

800-541-2593

www.questsoftware.com

Rebus Systems

Secaucus, N.J.

401-247-4925

www.rebusgroup.com

ReClaim Technologies

Newark, Ohio

740-344-6956

www.reclaimtech.com

Red Oak E-Commerce Solutions

Red Oak, Va.

512-363-4297

www.roesinc.com

Results International Systems

Dublin, Ohio

800-875-2126

www.resultscorp.com

RGL

Englewood, Colo.

303-721-8898

www.rgl.com

Risk Sciences Group, Inc.

Schaumburg, Ill.

800-619-0224

www.risksciencesgroup.com

Robert E. Nolan Co.

Dallas, Tex.

972-248-3727

www.renolan.com

SAP America

Newtown Square, Pa.

888-727-8591

www.sap.com

Sapiens Americas Corp.

Cary, N.C.

919-405-1510

www.sapiens.com

Scene Genesis

Pittsford, N.Y.

585-387-9883

www.scenegenesis.com

SeaTech Consulting Group, Inc.

Torrance, Calif.

310-328-8119

www.seatech.com

Simsol Software

Orlando, Fla.

800-447-4676

www.simsol.com

Sirius Financial Systems, Inc.

Englewood, Colo.

303-209-5914

www.sirius-inc.com

Stargate Solutions Group

Indianapolis, Ind.

317-469-7512

www.stargatesolutions.com

StarTech Software Systems

Novi, Mich.

248-344-2266

www.startechsoftware.com

SunGard Sherwood Systems

Armonk, N.Y.

877-777-8437

www.sungard.com

Tropics Software Technologies, Inc.

Sarasota, Fla.

941-955-1234

www.gotropics.com

Trumbull

Windsor, Conn.

877-285-2174

www.trumbull-services.com

Valley Oak Systems

San Ramon, Calif.

925-242-4600

www.valleyoak.com

Visibillity, Inc.

Chicago, Ill.

888-484-7424

www.visibillity.com

VS Visual Statement, Inc.

Kamloops, B.C., Canada

888-828-0383

www.visualstatement.com

XACT TeleSolutions

Mt. Juliet, Tenn.

800-558-1545

www.xactservices.com

Xactware

Orem, Utah

800-424-9228

www.xactware.com

XDimensional Technologies

Brea, Calif.

800-789-2567

www.xdimensional.com