Measuring and estimating the costs and extent of insurance fraud is paramount to raising our understanding of the insurance fraud phenomenon. Such estimates help insurers to prioritize problem areas and to efficiently allocate their resources in the fight against insurance fraud. Without reliable estimates the insurance industry cannot make credible statements to the public regarding insurance fraud and consequently raise the general awareness of the problem.
Can one measure fraud?
The first serious attempts to measure the costs and extent of insurance fraud were made only in the second half of the eighties of the previous century. Due to increased competition in the insurance services market and desire to reduce the costs, these attempts became more frequent in the nineties.
Regardless of all the efforts, it is still very difficult to accurately estimate the extent of insurance fraud or to quantify it. Viaene and Dedene cite several reasons for this (Viaene and Dedene, 2004). Firstly, they emphasize that the nature of insurance fraud is covert and thus difficult to estimate. Moreover, insurance fraud is as dynamic as the business environment. Perpetrators make use of every opportunity presented to them. Therefore, the estimated gap between detected and undetected fraud is rather considerable.
A major obstacle in measuring insurance fraud is the absence of a measuring standard that would be widely accepted in the insurance industry. This is supported by a study in which four claim reviewers were asked to review the same claim files and identify the suspicious ones (Derrig and Ostaszewski, 1995). While all claim reviewers assessed 5 – 10 percent of the claims as suspicious, no claim was assessed as suspicious by all four reviewers. The data on insurance fraud and the estimate of the extent of the insurance fraud phenomenon hence need to be interpreted cautiously.
As regards the estimate of the extent of insurance fraud, the paper by Sharon Tennyson provides interesting insights (Tennyson, 2008). The findings of the studies described in the paper defy the generally accepted estimate that insurance fraud constitutes 10 per cent of all claims. Namely, the studies have shown that the elements of insurance fraud were present in one quarter to one third of claims, whereas insurance fraud could be proved in only 3 to 5 per cent of all the reviewed claims.
These numbers correspond to my experiences, although more in terms of value than in number of claims, which may be a huge difference. When implementing anti-fraud systems with various non-life insurers, we would usually set the confirmed frauds target at 4 to 5 per cent of claims in value. This number, however, would not include numerous fraudulent claims that were detected and rejected by claim assessors themselves, without involvement of the anti-fraud unit.
The cost of insurance fraud is indicated in further studies presented by Tennyson. Apparently, the average costs of car repair claims are higher by 32.5 per cent when the repair is covered by the insurance company in comparison to the repairs without such coverage. Another study has shown that the number of claims for bodily injuries which are difficult to prove medically is higher by 42 per cent in those countries where lump sum compensation claims for bodily injuries are allowed and are not limited to medically proven injuries. The deductibles may also induce insurance fraud. Namely, a higher deductible resulted in 25 per cent higher car damage claims in value in cases where only one vehicle was involved, while in cases with more vehicles and thus more witnesses, there was no such substantive increase in the value of claims.
Viaene, S., Dedene, G.: Insurance Fraud: Issues and Challenges. The Geneva Papers on Risk and Insurance, vol. 29 No. 2, April 2004, pp. 313–333.
Derrig, R. A., Ostaszewski, K. M.: Fuzzy Techniques of Pattern Recognition in Risk and Claim Classification. Journal of Risk and Insurance 62 (1995), pp. 447–482.
Tennyson, S.: Moral, Social and Economic Dimensions of Insurance Claims Fraud. Social research Vol. 75: No 4, 2008, pp. 1181–1204.