Inspection Scheme Selecting And Evolutionary Game Analyzing For Insurance Frauds

Insurance fraud is a big challenge for the insurance industry, about 20%-30% of claims are fraud, while the inspection effects can hardly reaches 100% because of moral hazard and asymmetric information. So how to choose the right investigate strategies for fraud claims is a hard and important question for the insurance company.This study mainly contains two parts: the optimal inspection strategy in static situation and an analysis of the behaviors of the customers and the company based on the evolutionary game theory.For the optimal inspection strategies, we compared three different inspection strategies. In reality, the insurers tend to use a double-stage inspection strategy, which means the insurers first audit all the claims, select the doubtful claims(which might be a fraud claim); and the second step is make a more careful scrutiny with the select claims to find out the real fraud. While there is an extreme condition that the company inspect all claims carefully without any screening. We begin our research from a very practical view that the insurers have two opposite objectives: the maximum profit and the maximum effect of investigate and we try to find out the optimal inspect strategy. This paper talked about an new strategy that similar with the tradition strategy but in the second step, the company inspect not only the selected claims but also a random part of the claims.And the result shows that none of the three strategies works best in all situations, and the new strategy we designed can reach a goal which the traditional can’t get. At last we use give examples to explain the conclusion by using MATLAB.And then we use evolutionary game theory to analysis the changes of the behaviors of the company and the applicant. For the group of customers who have a very low moral cost, they have two strategies when no accident happens: cheat for a compensation or be honest doing nothing,and the company also have two strategies: inspect the claim or not. This chapter discussed two different kind of fraud: one is fraud without any accidents and the other is exaggerating the loss while an accident is really happen. For the first type of fraud, we discuss a forwardly check strategy which means we do the exam while the customer doesn’t apply for the claim as well as the normal only exam the claims.For the second type of fraud we discussed a new strategy that give the honest customer a reward compare with the normal one that won’t give the customers any rewards who have a claim.And we find the results are quite similar while the new strategy can speed up the time to the ESS,at the same time the distinguish of the types of customers are different.The results of the fraud evolutionary game shows that the strategies of the players can reach to different results in different conditions such as the inspection level, the moral level, the strategies the company take unlike the traditional evolutionary game theory. And specially the strategies of the players, both the company and the customers may change their behaviors based on each other, and reach to a recurrent state.