You can search for articles in back issues of Contingencies from July/August 2000 to March/April 2009 using the search box to the right. Simply type in subject words, author's name, or article title and click search. To search for articles from May/June 2009 to the present, go to the current digital issue of the magazine and use the search function on the left of the top navigation bar.

Google Custom Search

A Hard Look at Soft Fraud

By Kevin Bingham, John Lucker, and Mo Masud

PDF version

It’s not only the criminal masterminds who cost insurers millions of dollars in fraudulent claims. What about those ordinary folks who see nothing wrong with cheating the insurance company—just a little?

ACROSS THE COUNTRY, states have been increasing their budgets for the prevention and prosecution of insurance fraud. In New Jersey, the state Office of the Insurance Fraud Prosecutor reported 2004 fraud-fighting activity that included more than 70 criminal indictments involving 135 defendants that resulted in more than $16 million in criminal fines and restitution. In New York, the Insurance Frauds Bureau executed 815 arrests during 2004, setting a new record. Arrests in New York have shown a year-to-year increase for more than a decade, rising by more than 480 percent since 1995. Massachusetts, California, and other states have also been actively leading the fight against insurance fraud.

On a weekly basis, it’s not uncommon to read stories about hard fraud, such as auto insurance crash rings, the illegal funneling of insurance company assets, employee embezzlement and bogus claim payments, submitting of phony bills to Medicare, slip-and-fall mills, “rent-a-patient” scams, arson, etc. With the fight against hard fraud beginning to show significant rewards in terms of prosecution, public awareness, and the lowering of insurance rates in some states, it’s important that we don’t ignore the impact of another form of fraud, known as soft fraud, that’s thought to have a dramatic impact on an organization’s losses and overall claims costs. In fact, some experts believe that soft fraud is a much larger problem than hard fraud.

Everybody Does It

Unlike the hard fraud examples discussed above, soft fraud is much harder to detect and address because it’s often the result of abuse of the insurance process, claim exaggeration, embellishment, or opportunistic claim filing situations.

Soft fraud isn’t necessarily the result of egregious criminal intent and is typically done by individuals without a clear criminal history or criminal profile. Whereas most people express moral outrage at hard fraud, recent polls show that more than 25 percent of respondents thought it acceptable to pad insurance claims. Furthermore, even more people felt it was all right to embellish claims in order to recover insurance deductibles, and a similar number of people felt it was reasonable to distort policy information to reduce the amount of premium they owe to insurance companies.

Specifically for workers’ compensation claims, where fraud is a widespread problem, nearly one-third of Americans surveyed said it’s OK for employees to stay home from work and receive workers’ compensation benefits because they feel pain, even though their doctors could certify them as ready to return to work. One in five workers stated that they know of fraud in the workplace, but nearly 30 percent wouldn’t report insurance abuse committed by someone they knew.

The statistics paint a challenging picture of the environment insurance companies face when trying to control soft fraud. Whereas hard fraud usually represents criminal activity, something most consumers are strongly against, it’s surprising the number of individuals who feel that there’s nothing wrong with soft-fraud-related behavior. In reality, consumers often accept soft fraud as a fact of life and in some cases view such insurance system abuse as an entitlement to balance out perceived inequities in the insurance system.

Therefore, insurance companies must not only devise a method to detect soft fraud; they must also implement policies and interventions designed to modify customer behavior and improve perceptions about the acceptance of such abusive behaviors, such as padding insurance claims.

Whereas most people express moral outrage at hard fraud, recent polls show that more than 25 percent of respondents thought it acceptable to pad insurance claims.

Blank Slate

The fight against soft fraud is in its early stages, with very little written on the topic. Unlike the battle against hard fraud, the battle against soft fraud isn’t as precise and has the potential to lead to more false positives (i.e., claims modeling indicates fraudulent activity where none exists).

The potential number of false positives that emerge from some analytical solutions may represent an imperfect solution that can be intellectually uncomfortable for some actuaries, legal staff, and insurance executives. Yet, an imperfect solution that’s balanced with careful business rules and low-impact operational processes that are non-confrontational and non-intrusive can offer immense value to an insurance company. Such approaches, combined with the law of large numbers, can provide valuable insight into consumer behavior and some of the reasons that people do certain things. Equally important is the financial benefit to insurers, as they have the potential to reduce overall claims costs by as much as 5 percent through a broad soft fraud management strategy.

In order to address the complications of modeling and detecting soft fraud, it’s recommended that targeted claims management be employed to increase the touch points with the insured. By increasing “customer service” (i.e., proactive calls for assistance, requests for supporting claim information, etc.), the insurance company is letting customers know that the insurer is paying attention to claims being submitted and how the claims proceed through the management and settlement process.

Implementing a policy to record all customer phone conversations when reporting claims or providing additional information is an example of using a gentle technique to monitor customer behavior and potential fraudulent activity. When customers know their conversations are being recorded, they’re less likely to embellish a claim and report incorrect information to the insurer. The use of such non-confrontational and non-intrusive techniques can work without customer alienation or increased complaints.

Furthermore, a case can be made that improved customer service on “high touch” claims can actually reduce the number of soft fraud cases because it’s less likely customers will take advantage of a company that’s treating them fairly.

Predictive modeling applications for soft fraud include targeting claims duration and severity management in order to identify potential claim abuse. Early identification helps the insurer to better allocate customer service and claims triage resources to claims with the highest potential for fraud, while at the same time balancing the need to maintain quality of care and service to all customers.

Lots of Data

How is all of this accomplished? It starts with data, lots and lots of data. Valuable data buried in an organization’s claims, case management, and policy and loss control systems contain important information about each claim and the individual claimant’s characteristics where a greater propensity for embellishment exists.

In addition, organizations can leverage various external data sources, such as the Bureau of Labor Statistics, the U.S. Census, AMA physician licensing data, Pressley Reed, Med Stat, and others, to gain additional insight on behavioral, demographic, and environmental factors. External data sources are also valuable in benchmarking the effectiveness of an organization’s existing claims management and triage process by comparing utilization statistics against industry duration and severity norms.

Organizations that explore the use of predictive modeling applications often raise the question of data completeness and data quality as a potential inhibitor to a successful predictive modeling project. Our response is to remind organizations that predictive modeling applications aren’t accounting or actuarial software packages requiring 100 percent data reconciliation. These applications examine patterns in the raw data to draw insightful conclusions. Even if 80 percent of a data variable is complete, that’s often more than enough to draw a statistically significant conclusion about its predictive power.

These patterns in the data are carefully analyzed by applying powerful statistical algorithms and predictive modeling techniques. In the case of soft fraud detection for workers’ compensation claims duration and severity, a behavioral/environmental score or a relative claims duration and severity index would be generated for each claim, in addition to a medical complexity score. These individual scores would highlight potential areas of concern relative to industry norms and other claims for similar injuries, job classification, etc.

It’s important to note that these modeling applications should be viewed as a business process solution and not an IT solution. This is where a business rules engine plays a major role in claims management.

Predictive modeling scores are meaningless unless specific business actions are taken on claims with varying degrees of soft fraud potential. Again, in a disability and workers’ compensation context, the business rules engine will examine the clinical component and behavioral/environmental score and recommend specific claims management and triage actions that result in reduced time off from work. These actions could result in a recommendation of specific outreach or intervention, automatic assignment of the case to an advanced case management unit, recommendation of the claim for SIU audit, or automatic payment of a claim.

The end result is a powerful analytical tool that helps identify potential areas of concern that otherwise may have gone undetected and resulted in higher claim costs.

Although the battle is well under way against hard fraud, the battle against soft fraud is just beginning. By recognizing that soft fraud is a different problem with different underlying symptoms, executives can leverage predictive modeling tools to ensure that each claimant knows that the insurer is actively monitoring the status of each claim. With between 11 cents and 30 cents of every dollar being lost to soft fraud, insurers must devise strategies to control soft fraud as part of their overall operational strategy. Predictive modeling may not be 100 percent accurate, but significant value and financial savings can be gained from an imperfect solution that helps organizations tackle this growing problem without alienating their customer base.

KEVIN M. BINGHAM is a senior manager at Deloitte Consulting LLP in Hartford, Conn.; John Lucker is a principal at Deloitte Consulting LLP in Hartford and National Co-Leader of their Advanced Quantitative Services Practice (data mining and predictive modeling); Mo Masud is a manager at Deloitte Consulting LLP in Hartford, also in the Advanced Quantitative Services Practice.


(1) Annual Report of the New Jersey Office of the Insurance Fraud Prosecutor. Trenton, NJ: March 1, 2005. Online. New York State Insurance Frauds Bureau Annual Report to the Governor. New York, NY: 2004. Online. Accenture. Insurance Fraud Study. New York, NY : 2003. Online Insurance Research Council. Public Attitude Monitor, Workers’ Compensation Fraud. Malverne, PA: 1999. Online. Insurance Research Council. Insurance Services Office, Malverne, PA: 2002. Online.


Contingencies (ISSN 1048-9851) is published by the American Academy of Actuaries, 1100 17th St. NW, 7th floor, Washington, DC 20036. The basic annual subscription rate is included in Academy dues. The nonmember rate is $24. Periodicals postage paid at Washington, DC, and at additional mailing offices. BPA circulation audited.

This article may not be reproduced in whole or in part without written permission of the publisher. Opinions expressed in signed articles are those of the author and do not necessarily reflect official policy of the American Academy of Actuaries.

March/April 2006

A Hard Look at Soft Fraud

New Catastrophe Models for Hard Times

The Value of Human Life

In the Eye of the Beholder

Inside Track:
Nosy Data


Making Good Sausage

Up To Code:
Living With Precept 10

Policy Briefing:
Where Policy Meets Politics

Health-Related Quality of Life

The Future of Historic Studies

Statistical Miscellany:
2005 A Record Year for Casualty Claims

Peculiar Star Position

Second-Order Effects

2006 Directory of Actuarial Software

Past Issues

Contact us

American Academy of Actuaries