Insurance fraud has been a problem for insurance companies for centuries. According to different estimates, insurance companies are losing between 3% and 18% of claims expenditures to fraud, which exceeds 100 billion euros annually. Computer technology has been identified to have a crucial role in insurance fraud management. Fraud management system is a solution that supports all fraud management process activities: detection, prioritization, investigation, mitigation, redress, sanction, prevention, deterrence, continuous improvement and process monitoring.
Fraud appear in different industries, e.g. telecommunications, healthcare and health insurance, banks, credit cards, e-business, identity theft, tax fraud and tax evasion and last but not least motor insurance. There is a plethora of insurance fraud, and what is more, fraudsters are always inventing new types of fraud, forcing insurance companies to face the problem in a proactive manner.
In literature we can find numerous researches on components to support various fraud management activities, yet, only a few researchers demonstrated how such a component affects the business level performance. Thus proposed components, supporting the same fraud management activities, are incomparable. Moreover, it is not clear, whether certain proposed components even affect the business performance.
There is a clear need to establish a clear connection between fraud management system components and key performance indicators on the business level. In the dissertation, we propose a method for fraud management systems design based on business performance metrics. Method can be used to identify the most suitable fraud management system components, provided the set of KPIs that the insurance company wants to improve.
The research was conducted in three phases: method construction, correlation setup based on the information from Slovene insurance companies, and method evaluation on three concrete case studies. In different phases we employed different research methods, such as literature review, expert interviews and case studies.
Method bases on the business process metamodel and is comprised of five artifacts, namely (1) set of key performance indicators, (2) correlation matrix between KPIs and activities, (3) set of fraud management process activities, (4) correlation matrix between activities and FMS components, and (5) set of fraud management system components. Using these five artifacts, one can, based on a set of KPIs an insurance company would like to improve, select the most appropriate fraud management system components to do it. By using inverse method, one can predict the impact of the implementation of FMS components on business level performance.
Method was evaluated on three case studies. In each case study we first identified key performance indicators, a insurance company should improve. Based on these KPIs, we identified the fraud management system components. Then we predicted the impact of the implementation of these FMS components on business performance. Last, one year after the implementation, we again measured the KPIs and compared the new performance to the goals and prediction before the implementation.
In all three case studies we observed a positive outcome and a proof that the method works on the Slovene market. All three case studies resulted in a positive change of the KPIs, targeted to improve. Moreover in all case studies an improvement of general performance was observed. There was a increase in total savings (increase between 18.74% and 42.67%) and profitability (increase between 34.81% and 83.94%).
We wish to test the method outside of Slovenia and set up an international benchmark based on enhanced data set. The benchmark will enable insurance companies to easily compare business performance. Furthermore, we hope to extend the method to enable the identification of the most suitable fraud management process changes in order to improve fraud management performance.
|
