Effectiveness of Insurance Fraud Interventions - Literature review Example

The fraud group consisted of 917 claims, 689 men and 228 women, while the control group comprised 644,939 claims, 423,976 men, and 220,968 women. Separation of men and women was based on the fact that sex differences yield different behaviors. While this may be an effective approach, it is quite challenging and risky especially considering individuals who make claims over the phone and a decision has to be made within a short time frame. Viaene et al. also employs the automatic detection approach that uses claim screening systems to identify fraud in insurance claims.

The research used data sets from a Spanish insurer of cars from accidents comprising of 2403 claims made in the year 2000.2 Although there were no control groups, the claims would be classified into two groups’ coded fraudulent claims and honest claims.Conversely, Ai, and Brockett, Golden explore a method of classifying insurance claims according to their suspiciousness of fraud and non-fraud.3 More specifically, the paper uses the PRIDIT (Relative to an Identified Distribution Integral Transform) analysis to assign numerical scores on the predictor variables.

Some of the predictor variables used include evaluation of time between incident occurrence and claim filling and age of the claimant. The paper used data sets provided by the Automobile Insurance Bureau (AIB), Massachusetts comprising of 1399 personal injury protection claims for automobiles hence there was no use of control groups. Guillen et al. use the PRIDIT-FRE (PRIDIT-based Fraud Rate Estimation) method to assess suspicion in the insurance claims presented without the use of a control group.

The PIDIT-FRE method uses the same underlying mathematical algorithm used in the RIDIT approach but its main purpose is fraud estimation.