Research On Fraud Risk Management In The New Rural Cooperative Medical Insurance

Medical insurance fraud is a common and serious problem which has drawn wide attention in all countries. In China, medical insurance fraud cases occur frequently with the implementation of the New Rural Cooperative Medical System (NRCMS). Fraud poses great threat to the safety of NRCMS fund and the sustainable development. How to prevent and curb medical fraud has become a major task that the government must tackle because it closely relates to the implementation of preferential policies for agriculture, farmers and rural areas. Therefore, the research on medical fraud is of practical significance and applied value. This dissertation, based on the economic analysis on the formation mechanism of the new rural cooperative medical fraud and risk management theory, focuses on the analysis, detection, measurement, prevention and curb of NRCMS fraud. It employs comparative and interdisciplinary research method combined with normative analysis and empirical research, qualitative and quantitative analysis methods. Both the research approaches and focuses complement the existing achievements in this field and are of certain academic value.The main contributions of this research lie in the following three aspects. First, it investigates the current situation, formation mechanism of NRCMS fraud and fund management problems. Specifically, fraud risk factors and characteristics, fraud principals, means, basic types are summed up based on the analysis of some fraud cases in line with the NRCMS implementation procedure. The formation mechanism of NRCM fraud was elaborated from perspectives of game theory, principal-agent theory and criminal economics. This provides theoretical basis for the follow-up study.The analysis shows that the medical fraud subjects range from occupational fraud gangs, designated health care institutions (doctors), insured farmers and NRCMS management staff with the first two causing the greatest loss (80%) to the fund. With the reform of the NRCMS reimbursement system, the insured farmers’medical expenses can be reimbursed immediately at hospitals where they get medical services. The new system will effectively curb dirty transactions made by the occupational fraud gangs, insured farmers and NRCMS management staff. However, under the new system, designated hospitals (doctors) tend to commit frauds and abuse by their monopolistic and professional advantages. Therefore, it’s a critical to supervise designated hospitals (doctors) to combat frauds. Analysis also shows that fraud detection rate, investigation cost, penalty cost and asymmetric information affect fraud perpetrator’s decisions, providing a constructive approach to the establishment of anti-fraud system. In addition, China’s anti-fraud under the NRCMS has limitations in legal system, supervision system, professionals, anti-fraud techniques and supervision over designated medical institutions.Secondly, the fraud detection model based on BP neural network has been set up to facilitate the audit of fraud risks. The instant reimbursement system has been carried out in NRCMS agencies in 90% counties. The inter-provincial long-distance reimbursement system will be initially implemented in 2015. So, Designated hospitals’fraud and abuse become the greatest threaten to the safety of NRCMS fund. Therefore, to identify frauds of designated medical institutions becomes the key to the research of fraud detection system. Specifically,14 indexes from medical service indicators, medical care expense indicators and medical compensation indicators are selected to make up the designated hospital fraud detection system such as’total inpatients".’average hospitalization expenses growth rate’ and’the actual compensation rate". An empirical study on a county’s medical insurance fraud data in 2012 is made on the basis of the newly-constructed BP neural network fraud detection model (3-11-1) and dimensionality reduction through principal component analysis approach in comparison with the performance of logistic model. The empirical study reveals that BP neural network identifies 100% fraud and legal samples for training, 80% fraud samples and 78.95% legal samples for testing. Its total recognition rate reaches 79.31%, representing superior performance over Logistic model. The logistic model can only identify 43.8% and 50% fraud samples for training and testing respectively, basically without any value for detection.Thirdly, the measurement of medical fraud risks are explored with single distribution of loss distribution approach, POT model of extreme value theory, and piecewise defined loss intensity distribution of the loss distribution approach (PSD-LDA), combined with VaR and TVaR methods, measure new fraud risk loss value. The fraud loss data released by the media during 2004-2012 has been investigated for empirical research. The results show that the PSD-LDA method is the best among the three models. The PSD-LDA method combines the advantages of both, considering the characteristics of medical fraud risk "low frequency with high loss" and "high frequency with low loss", a more accurate measurement of risk of the NRCMS fraud. Therefore, it is more reasonable to study fraud risk with the PSD-LDA method. The lognormal distribution, the generalized Pareto distribution fitting fraud risk loss distribution and the Poisson distribution for the loss frequency distribution are employed to caculate VaR and TVaR value of the risk of fraud through Monte Carlo simulation with the new fraud risk reserve fund of 635,797,000 Yuan and net premium loss of 8,549,000 Yuan. In addition, the paper also discusses the application of VaR or TVaR into fraud risk precaution management.Finally, the dissertation establishes the anti-fraud system. The successful experience in USA medical insurance fraud has been surveyed for this purpose. USA has achieved remarkable fruits in combating medical frauds through the complete anti-fraud law enforcement system, the anti-fraud legal system and series of anti- fraud regulations, methods and measures owing to its long-term experience in fighting against medical frauds. The anti-fraud system has been set up on systematic and technical levels with reference to the American successful anti-fraud experience, the analysis of NRCMS fraud risk formation mechanism, combined with the reality of the country. The system level mainly includes the development of anti-fraud law system, anti-fraud law enforcement agencies, the release system, the establishment of clinical pathway management system, and the internal control system and supervision mechanism to establish the third party insurance audit. The technical level means the development of fraud risk management system, including fraud risk analysis, risk assessment and monitoring of precaution subsystem. The establishment of anti-fraud system serves an important institutional guarantee and effective technical means to tackle frauds.