Insurance Fraud: Expert Review - Coursework Example

INSURANCE FRAUD: EXPERT REVIEW by Discussion The research identifies several challenges concerning fraud experts and their qualifications required for position. There are several instances which officially educate fraud experts; they include The Association of Certified Fraud Examiners (ACFE), the Institute of Counter Fraud Specialists (ICFS) and Accredited Counter Fraud Specialists (ACFS) in the UK. Even though theoretical education exists, most fraud expects are educated on the basis of their companies. It is a regular practice to make up a crib sheet which outlines basic fraud indicators; according to Belhadji et al. (2000), it contains only frequent characteristics of fraudulent cases while minor fraud indicators are ignored. Lack of training and over-reliance on fraud expert work experience result in low efficacy of fraud detection. If to consider that fraudsters tend to generate new ways to commit fraud quickly (Šubelj, Furlan & Bajec, 2011) and the mean chance to detect fraud is only 54% (Bond & DePaulo, 2006; Vrij, 2008a), the research which examines attitudes and beliefs held by fraud experts is crucial in order to improve the efficacy of insurance fraud detection. The sample analysis of 46 fraud experts shows that in some ways their decisions are predetermined by common beliefs and stereotypes which does not let them judge objectively. First thing to notice is the lack of formal definition of fraud expert job position. The sample enrols people who occupy different positions but share the same responsibilities in their companies with the most common titles fraud investigator (33%) and fraud manager (19%). In average, fraud experts are experienced in their area with the mean result of 2 years or more. The absence of information about education of experts in the sample serves as a limitation for this research which can be covered by further researching of fraud expert efficacy and level of education. The results of this research are applied to the results of insurance fraud demographics study from the previous chapter. The comparison of beliefs help by experts and actual fraudsters profiles immediately shows several inconsistencies. For instance, one of the core assumptions concerning male perpetrators is supported by the majority of participants of the study. The results of motor fraud correspond to the results gathered in the study. 63% of all participants expected 50-75% of all motor fraud cases to be committed by men. The mean result of the fraud demographics research shows that 62% of all fraud crimes are committed by men. In case of household insurance fraud, the assumption about gender of the offender can be misleading. 43.6% of all analysed household fraud cases are committed by women.  Most professional fraud detectors are more attentive working with men. Stereotypical beliefs about female behaviour decrease chances to detect motor insurance fraud cases committed by females. The expected age of offenders is also different from the actual demographic profiles. Fraud detectors tend to think that fraudsters are 5-6 younger that they actually are. People in their thirties are more likely to experience more rigorous interview than older policy holders who commit the majority of fraud cases in insurance industry. Misleading beliefs about young offenders can decrease attention of professional fraud detectors in cases which involve people in their forties and seniors. Prior studies indicate that people with prior claims or criminal records are more likely to display fraudulent behaviour (The Home Office, 2003; Berman & Dar, 2013). In fact, 69.7% out of 78 claimants in household insurance fraud and 56% claimants in motor insurance fraud have prior claims. The frequency of prior claims is treated as a factors contributing to fraud insurance risk group definition. Opinions of insurance fraud expects do not fully coincide with the results of the research. Most of them expect at least 1 (19.6%) or 2 (23.9%) prior claims from motor insurance fraudsters. Only 4.3% of respondents believe that claimants with no prior claims can be suspected in fraud. Evidently, it creates a certain trustful attitude to people with no prior claim which reduces chances to detect fraud. In the analysed sample of motor car fraudsters, 34% of motor insurance fraud cases have no prior claim history; moreover, 20% of fraudsters conceal the information about their prior claims. In this way, the absence of prior claims in the case cannot serve as a vindicating argument for the fraudster. Beliefs about marital status of fraudster also display discrepancies compared to the real cases of insurance fraud. Only 13% out of 78 household claimants were single with the overwhelming majority of married (60.3%) and a big group of cohabiting (21.9%) fraudsters. Motor claimants have very close results for single and married status (41.2% and 39.2% accordingly). The results of questionnaire with fraud detectors show that they are influenced by the common belief about marital status of fraudsters. 46% of inquired professional lie detectors expected motor fraudster to be single. In case of household claimants, they expected that up to 50% of claimant would be single. Geographically, most experts expect fraud to be committed in highly urbanised areas with London on the top (22.7%). Contrary to this, West Midlands and West Yorkshire top the list of household fraudulent claims with 7.7% each. Only 3.4% of all household cases examined in the study happened in London. Motor fraud is more typical for big cities as 14% of all cases take place in London. Still, beliefs about other regions negatively influence experts ability to detect fraud in rural territories or smaller cities. The cases of loss and claim disclosure rates coincide as they fully depend on work experience of each fraud investigator involved in the study. The analysis of reasons which motivate these beliefs most experts admitted their work experience (28.3%) and claim experience (17.4%) with no reference to professional education and training. These results show that people depreciate educational courses in the sphere pf criminal psychology and prefer to rely on their experience in lie detection. The lack of professional training has been already noted by other researchers (Bond & DePaulo, 2006; Vrij, 2008a). In this way, the findings of this research are persuasive enough as a reason to create sustainable training program based on complex analysis of criminal demographics in the sphere of household and motor insurance. The implication of the results of this research in fraud detection training is critical for the insurance industry. The main challenge is that companies tend to spend as little money as possible on case investigations and corporate training for their employees (Derrig, 2002). Companies fail to conduct their internal research to test the effectiveness of their lie detection methods and tend to rely on already existing policies recommended by prior (out-of-date) studies. For example, Artís, Ayuso & Guillén (2002) promote the use of binary logical models to identify risks in cases of motor fraud. The application of this model heavily depends on personal experience of the investigator. It has been already noted that over-reliance on work experience leads to overconfidence in own ability to detect fraud (Baker et al., 2013). Obviously from the discrepancies between the beliefs about fraudsters and their real profiles, this over-confidence is unmotivated. As well as the previous part of the research, this chapter identifies the severe lack of database available for insurance lie detectors. The creation of local and nation-wide insurance fraud data bases is required to make it possible to improve the industry by the application of the latest IT interventions. According to Phua, Lee, Smith & Gayler (2010), there are different supervised, partially supervised and unsupervised automated digital fraud detection intervention based on data-mining technique but all of them require constant availability of digital data bases. These techniques have been successfully applied to detect audit fraud because it retains high level of objectivity in research. The use of IT interventions depends on the availability of comprehensive data bases and chosen algorithm of fraud detection. The algorithm is constructed on the basis of selection of the most typical characteristics of fraudsters that attribute them to the crime risk group. Insurance industry in the UK still relies on professional lie detectors and disregard costly but effective technological interventions. A number of studies show that they can improve the effectiveness of lie prediction and save time spent on case analysis and investigation. According to Liu & Chen (2012), application of data mining technique to detect insurance fraud results in visible enhancements in prediction probability. Their study of GA-Kmeans and MPSO-Kmeans algorithms in automated insurance fraud prediction based on two cases resulted in 86%-93% of accuracy. All demographic variables discussed in this research correspond to variable used for algorithm adaptation to local specific trends in insurance fraud. The next implication of this research is to educate professional lie detectors. The research shows that they rely on their personal work experience as a primary source of information. Their experiences are different due to the sphere of insurance fraud and their geographic location. Moreover, common beliefs are influenced by gender and age stereotypes related to the demographic profile of the offender. Their knowledge supported with automated prediction methods can achieve great result. As far as algorithms are based on general information, the results of their calculations are required to be interpreted to make them more specific. Based on the results of this research, it is relevant to suggest that training of professional lie detectors needs to focus on two main aspects. The first aspect is objective critical thinking skill development. Stereotypes together with beliefs which develop as a result of work experience make fraud experts less attentive to people who do not coincide with their image of a typical fraudster. According to Perri & Brody (2012), fraudsters can resort to various techniques to mislead investigators; they can use projected bias technique appealing to the common values they share with insurance agent or fraud investigator. Emotional appeals remain their strongest tool applied to neutralize skepticism of fraud investigators. Objectivity and emotional coolness are required to prepare lie detectors to be careful with every client. The second aspect is to teach professional lie detectors to use technologies in the process of investigation. The reference to data base to retrieve prior claims is not enough to improve efficacy of lie detection. Utilization of advanced IT technologies is required to provide more accurate investigation results. Regular training programs need to be organized for all employees to teach them how to work with new algorithm, Perhaps, insurance companies need to reconsider their economical approach to fraud investigation and spend more on education for their employees. The research identifies that fraudsters change with time, but investigators stay the same. The scope of the study represents a massive data set for utilization for professional lie detectors on-site training program development. The data gathered in the process of issue examination can serve as a general crib sheet for all employees in insurance business. This study stimulates implementation of IT technologies in the UK insurance firms and encourages employer to pay more attention to professional development of their employees. Development of training program is required to educate investigators about current trends in insurance fraud which do not correlate with their professional experience. Additional education for fraud investigators is predetermined by the obvious influence of common stereotypes related to age, gender, prior record and location of insurance fraud. Finally, this research comprehensively represents all major inconsistencies in insurance fraud detection; it can serve as a guide for insurance fraud investigators in their daily work. References Artís, M., Ayuso, M., & Guillén, M. 2002. Detection of automobile insurance fraud with discrete choice models and misclassified claims. Journal of Risk and Insurance, 69(3), 325-340. Derrig, R. A. 2002. Insurance fraud. Journal of Risk and Insurance, 69(3), 271-287. Liu, J. L., & Chen, C. L. (2012). Application of Evolutionary Data Mining Algorithms to Insurance Fraud Prediction. In 2012 4th International Conference on Machine Learning and Computing (ICMLC 2012), pp.-, Hong Kong. Perri, F.S. , & Brody, R.G., 2012. The optics of fraud: affiliations that enhance offender credibility. Journal of Financial Crime, 19(3), 305-320. Phua, C., Lee, V., Smith, K., & Gayler, R. 2010. A comprehensive survey of data mining-based fraud detection research. arXiv preprint arXiv:1009.6119. Read More