How do we Measure our Fraud Problem??
No, this is not a contemporary backgammon set, but instead a measurement tool that I found tucked away at a castle in Switzerland. Why it this significant to us in counter fraud? Based on the research currently in circulation, my specific research as faculty for Colorado State Global, and consultations with dozens of companies, one of the most significant challenges facing our industry is the relative lack of awareness of the fraud problem. That is, there is no universal or consistent measurement of fraud, which makes it difficult to accurately measure and report on the problem. As an industry, on a global level, we do not have one specific agency that collects fraud data; each insurance company, state or federal agency, or any other organization must devise its own reporting protocol.
There are many companies that measure fraud by single occurrence; some measure by prosecution, some by exposure, and others by dollars collected, and the list continues. In an informal experiment I performed, I asked four auditors to review the same exact claim files for potential fraud. The reviewers were consistent with their percentage of identification; each identified 7 to 12 percent as suspicious, yet not one of the same claims was labeled as fraudulent by all four auditors! In other words, they all picked different claims as having fraudulent indicators. There also appears to be an interesting ripple effect when considering the relative efforts of fraud identification. If a company invests very little time and effort into detecting and attempting to quantify fraud, then little fraud will be uncovered, providing a false sense of the problem. If preventative strategies are based on a meager detection approach, then the company will undoubtedly have a false sense of security due to low detection results, which could be quite contrary to its true fraud exposure.
To illustrate this inconsistency, the following are several examples of data and measurement disparity based on my consultations. Company A measured fraud based on dollars saved; that is, its annual fraud result was measured by the total number of dollars saved as a result of its fraud investigations. When further inquiry into this monetary amount was requested, it was discovered that the measurement appeared to be based not on the amount that the fraud investigation actually saved, or recovered, but on the reserve of the claim file at the time it was transferred into the fraud unit. If claim 123 was transferred into the fraud investigations unit with a reserve of $50,000 aggregate for all exposures, then this fraud department took a savings of $50,000 for this file, regardless of the outcome of the actual investigation. Another company (B) would measure its fraud rate by the percentage of cases where fraud was proved, that is, claims that it can deny. Company C would measure its fraud rate by the number of cases that enter into the unit, regardless of whether fraud is proven, or the claim simply sent back to the claims department for payment. As we can see, there are many different methods to measure and track fraud, and each company and agency seems to have a different method for this process, a method that creates inconsistent data and a false sense of the true nature of the problem. Two large companies based internationally add a unique perspective to this topic. Both made a conscious decision to refrain from measuring and sharing their fraud rate and accompanied savings, as they feel it gives the appearance that they take pride in denying their valued policyholders, a message they do not desire to disseminate.
The point of discussing measurement inconsistency in fraud data is not to create a doom and gloom perspective, it is to bring awareness to the fact that one of the major problems we have facing us in the fight against fraud is the lack of a universal definition. This lack of common ground makes it difficult for us to accurately communicate our problem to others within the industry and also to legislators and the political arena. Thus, one of the specific strategies we can start to ponder as fraud fighters is a way to create a benchmark measurement of fraud by creating common definitions and a formula for fraud identification. The challenge here also lies with the release of sensitive, confidential corporate information across companies; many organizations are hesitant to release specific information as they fear it may violate corporate policy.
In order to successfully develop an effective counter fraud policy, one must have the correct measurements in place. Quite simply, we must be measuring the right thing! As we have discussed, flexibility with counter fraud efforts is needed in order to adapt to changing fraud schemes, and as such, the measurement parameters also need to be modified routinely as well. If our measurements are focused on arson rings in the automobile line of business, but current trends are showing spikes in commercial home arson, then we must adjust our parameters accordingly. In addition, we must understand that in order to be accurate, our measurements must be consistent. Retrieving and reviewing data from only a few months prior will not provide the information needed to accurately assess our system of controls. We need to be sensitive to this and understand that oftentimes we need to participate in a cycle of data collection (which could be a year, or more) in order to gain meaningful data for interpretation.
The main message to be delivered in regard to measurement is to take a critical view of how data is collected and what specific parameters of measurement are used. Information can be skewed, and the resulting data will provide a different view of the direction for effective counter fraud efforts. Data is the source of all effective preventative efforts, but be guarded in how it is collected before application.