A rule-based fuzzy reasoning system for assessing the risk of management fraud

Detecting management fraud and assessing the risk of management fraud are significant issues confronting the auditing profession. Considerable theoretical and empirical research (Loebbecke, Eining, and Willingham, 1989; Bell, Szykowny, and Willingham, 1993; Fanning, Cogger, and Srivastava, 1995; and Hansen, McDonald, Messier, and Bell, 1996) has been accomplished investigating these issues. Building on this research, we demonstrate the construction of a rule-based fuzzy reasoning system to assess the risk of management fraud. The paper illustrates how fuzzy sets can be used intuitively to measure red flags on a categorical or interval scale, how different red flags can be combined using fuzzy rules, and how a single measure of the risk of management fraud can be derived. The knowledge base for this fuzzy reasoning system is developed by using the causal model of management fraud developed by Loebbecke, Eining and Willingham (1989), the empirical investigation of this model by Bell, Szykowny, and Willingham (1993), other researchers’ efforts and the authors’ judgments, using XpertRule software. The fuzzy reasoning system is tested using the fraud data provided by KPMG Peat Marwick. We discuss methods to magnify the knowledge base of this fuzzy reasoning system to make it a viable auditing tool, the costs and benefits of building a fuzzy reasoning system, and further extensions of this research. Copyright © 1998 John Wiley & Sons, Ltd.     

Bayesian Fraud Risk Formula for Financial Statement Audits

In this article we extend the work of Loebbecke et al. (1989 ) and illustrate the use of an evidential reasoning approach for developing fraud risk analysis models under the Bayesian framework. New formulations facilitating fraud risk assessments are needed because decision tree approaches previously used to develop analytical models are not appropriate in complex situations involving several interrelated variables. To demonstrate the evidential reasoning approach, a fraud risk assessment formula is derived and illustrated. The fraud risk formula captures the impact of the presence or absence of and interrelationships between the three ‘fraud triangle’ risk factors: Incentives, Attitude and Opportunities. The formula includes the impact of risks and controls related to these three fraud risk factors as well as the impact of forensic audit procedures and relevant analytical and other procedures that provide evidence for the presence or absence of fraud. This formula may be used in audit practice both to help plan the audit and to assess fraud risk sequentially as audit evidence is obtained.     

The Dempster‐Shafer Theory: An Introduction and Fraud Risk Assessment Illustration

This article introduces the Dempster‐Shafer theory (DS theory) of belief functions for managing uncertainties, specifically in the auditing and information systems domains. The use of DS theory is illustrated by deriving a fraud risk assessment formula for a simplified version of a model developed by Srivastava et al. (2007). In this formulation, fraud risk is the normalised product of four risks: risk that management has incentives to commit fraud; risk that management has opportunities to commit fraud; risk that management has an attitude to rationalise committing fraud; and risk that an auditor's special procedures will fail to detect fraud. The article demonstrates how to use such a model to plan for a financial audit where management fraud risk is assessed to be high. In addition, it discusses whether audit planning is better served by an integrated audit/fraud risk assessment as now suggested in SAS 107 (AICPA 2006a, see also ASA 200 in AUASB 2007) or by the approach illustrated here where a parallel, but separate, assessment is made of audit risk and fraud risk.     

财务舞弊公司董事会变更及其对审计师变更的影响——基于面板数据Logit模型的研究

本文以2002—2005年间因财务舞弊而被证监会、上交所和深交所谴责公告的61家A股上市公司为样本,以舞弊被公告前一年到公告后第三年为研究区间,运用配对样本T检验探讨后续治理中董事会成员和审计师如何变更,并建立面板数据Logit模型研究董事会变更对审计师变更的影响,得出了以下结论:从公告前一年到公告后第三年,财务舞弊公司的董事会发生了显著变更,但只有其非常规性变更才对审计师变更有显著正影响。然而,以控制组公司数据为对象,并未发现董事会变更会影响审计师变更。     

Why computer-mediated communication improves the effectiveness of fraud brainstorming

Current auditing standards require auditors to conduct a fraud brainstorming session aimed at considering ways in which the audit client's financial statements might be fraudulently misstated. Lynch et al. (2009) document that computer-mediated fraud brainstorming is significantly more effective than face-to-face brainstorming for generating relevant fraud risks. In this study, we code and analyze process-level data from the Lynch et al. (2009) study to understand the factors contributing to the greater effectiveness of electronic fraud brainstorming. Specifically, we conduct mediation analysis to discern the degree to which equality of participation and two measures of task focus contribute to greater fraud brainstorming effectiveness when using a computer-mediated communication system compared to traditional face-to-face brainstorming. We also examine participants' perceptions of ease of system use, satisfaction with the process, and satisfaction with the outcome. Overall, the results indicate that the primary reason for the greater effectiveness of electronic fraud brainstorming is the greater degree of task focus as revealed through the length of comments made when using computer-mediated fraud brainstorming. In an absolute sense, participants using electronic brainstorming felt that their brainstorming mode was easy to use and they were satisfied with the process and outcome. The primary contribution of this study is in enhancing our understanding of precisely why computer-mediated fraud brainstorming outperforms face-to-face fraud brainstorming.     

Insurance Fraud

Insurance fraud is a major problem in the United States at the beginning of the 21st century. It has no doubt existed wherever insurance policies are written, taking different forms to suit the economic time and coverage available. From the advent of “railway spine” in the 19th century to “trip and falls” and “whiplash” in the 20th century, individuals and groups have always been willing and able to file bogus claims. The term fraud carries the connotation that the activity is illegal with prosecution and sanctions as the threatened outcomes. The reality of current discourse is a much more expanded notion of fraud that covers many unnecessary, unwanted, and opportunistic manipulations of the system that fall short of criminal behavior. Those may be better suited to civil adjudicators or legislative reformers. This survey describes the range of these moral hazards arising from asymmetric information, especially in claiming behavior, and the steps taken to model the process and enhance detection and deterrence of fraud in its widest sense. The fundamental problem for insurers coping with both fraud and systemic abuse is to devise a mechanism that efficiently sorts claims into categories that require the acquisition of additional information at a cost. The five articles published in this issue of the Journal of Risk and Insurance advance our knowledge on several fronts. Measurement, detection, and deterrence of fraud are advanced through statistical models, intelligent technologies are applied to informative databases to provide for efficient claim sorts, and strategic analysis is applied to property‐liability and health insurance situations.     

Learning minimum variance discrete hedging directly from the market

Option hedging is a critical risk management problem in finance. In the Black–Scholes model, it has been recognized that computing a hedging position from the sensitivity of the calibrated model option value function is inadequate in minimizing variance of the option hedge risk, as it fails to capture the model parameter dependence on the underlying price (see e.g. Coleman et al., J. Risk, 2001, 5(6), 63–89; Hull and White, J. Bank. Finance, 2017, 82, 180–190). In this paper, we demonstrate that this issue can exist generally when determining hedging position from the sensitivity of the option function, either calibrated from a parametric model from current option prices or estimated nonparametricaly from historical option prices. Consequently, the sensitivity of the estimated model option function typically does not minimize variance of the hedge risk, even instantaneously. We propose a data-driven approach to directly learn a hedging function from the market data by minimizing variance of the local hedge risk. Using the S&P 500 index daily option data for more than a decade ending in August 2015, we show that the proposed method outperforms the parametric minimum variance hedging method proposed in Hull and White [J. Bank. Finance, 2017, 82, 180–190], as well as minimum variance hedging corrective techniques based on stochastic volatility or local volatility models. Furthermore, we show that the proposed approach achieves significant gain over the implied BS delta hedging for weekly and monthly hedging.     

Detecting Accounting Fraud in Publicly Traded U.S. Firms Using a Machine Learning Approach

We develop a state-of-the-art fraud prediction model using a machine learning approach. We demonstrate the value of combining domain knowledge and machine learning methods in model building. We select our model input based on existing accounting theories, but we differ from prior accounting research by using raw accounting numbers rather than financial ratios. We employ one of the most powerful machine learning methods, ensemble learning, rather than the commonly used method of logistic regression. To assess the performance of fraud prediction models, we introduce a new performance evaluation metric commonly used in ranking problems that is more appropriate for the fraud prediction task. Starting with an identical set of theory-motivated raw accounting numbers, we show that our new fraud prediction model outperforms two benchmark models by a large margin: the Dechow et al. logistic regression model based on financial ratios, and the Cecchini et al. support-vector-machine model with a financial kernel that maps raw accounting numbers into a broader set of ratios.     

Off-site monitoring systems for predicting bank underperformance: a comparison of neural networks,discriminant analysis,and professional human judgment

This study compares the ability of discriminant analysis, neural networks, and professional human judgment methodologies in predicting commercial bank underperformance. Experience from the banking crisis of the 1980s and early 1990s suggest that improved prediction models are needed for helping prevent bank failures and promoting economic stability. Our research seeks to address this issue by exploring new prediction model techniques and comparing them to existing approaches. When comparing the predictive ability of all three models, the neural network model shows slightly better predictive ability than that of the regulators. Both the neural network model and regulators significantly outperform the benchmark discriminant analysis model's accuracy. These findings suggest that neural networks show promise as an off-site surveillance methodology. Factoring in the relative costs of the different types of misclassifications from each model also indicates that neural network models are better predictors, particularly when weighting Type I errors more heavily. Further research with neural networks in this field should yield workable models that greatly enhance the ability of regulators and bankers to identify and address weaknesses in banks before they approach failure. Copyright © 2001 John Wiley & Sons, Ltd.     

Detecting insurance fraud using supervised and unsupervised machine learning

Fraud is a significant issue for insurance companies, generating much interest in machine learning solutions. Although supervised learning for insurance fraud detection has long been a research focus, unsupervised learning has rarely been studied in this context, and there remains insufficient evidence to guide the choice between these branches of machine learning for insurance fraud detection. Accordingly, this study evaluates supervised and unsupervised learning using proprietary insurance claim data. Furthermore, we conduct a field experiment in cooperation with an insurance company to investigate the performance of each approach in terms of identifying new fraudulent claims. We derive several important findings. Unsupervised learning, especially isolation forests, can successfully detect insurance fraud. Supervised learning also performs strongly, despite few labeled fraud cases. Interestingly, unsupervised and supervised learning detect new fraudulent claims based on different input information. Therefore, for implementation, we suggest understanding supervised and unsupervised methods as complements rather than substitutes.     

Predicting credit card fraud with Sarbanes‐Oxley assessments and Fama‐French risk factors

This research developed and tested machine learning models to predict significant credit card fraud in corporate systems using Sarbanes‐Oxley (SOX) reports, news reports of breaches and Fama‐French risk factors (FF). Exploratory analysis found that SOX information predicted several types of security breaches, with the strongest performance in predicting credit card fraud. A systematic tuning of hyperparamters for a suite of machine learning models, starting with a random forest, an extremely‐randomized forest, a random grid of gradient boosting machines (GBMs), a random grid of deep neural nets, a fixed grid of general linear models where assembled into two trained stacked ensemble models optimized for F1 performance; an ensemble that contained all the models, and an ensemble containing just the best performing model from each algorithm class. Tuned GBMs performed best under all conditions. Without FF, models yielded an AUC of 99.3% and closeness of the training and validation matrices confirm that the model is robust. The most important predictors were firm specific, as would be expected, since control weaknesses vary at the firm level. Audit firm fees were the most important non‐firm‐specific predictors. Adding FF to the model rendered perfect prediction (100%) in the trained confusion matrix and AUC of 99.8%. The most important predictors of credit card fraud were the FF coefficient for the High book‐to‐market ratio Minus Low factor. The second most influential variable was the year of reporting, and third most important was the Fama‐French 3‐factor model R² – together these described most of the variance in credit card fraud occurrence. In all cases the four major SOX specific opinions rendered by auditors and the signed SOX report had little predictive influence.     

Feasible fraud and auditing probabilities for insurance companies and policyholders

Insurance claims fraud is counted among the major concerns in the insurance industry, the reason being that excess payments due to fraudulent claims account for a large percentage of the total payments each year. We formulate optimization problems from the insurance company as well as the policyholder perspective based on a costly state verification approach. In this setting??while the policyholder observes his losses privately??the insurance company can decide to verify the truthfulness of incoming claims at some cost. We show simulation results illustrating the agreement range which is characterized by all valid fraud and auditing probability combinations both stakeholders are willing to accept. Furthermore, we present the impact of different valid probability combinations on the insurance company??s and the policyholder??s objective quantities and analyze the sensitivity of the agreement range with respect to a relevant input parameter. This contribution summarizes the major findings of a working paper written by Müller et?al. (Working Papers on Risk Management and Insurance (IVW-HSG), No. 92, 2011).     

Psychometric methods in accounting education: a review,some comments and implications for accounting education researchers

Psychometric instruments are widely used in accounting education research (Herring et al., Issues in Accounting Education, 4(1), 48–57, 1989; Benke and Street, Accounting Education: an international journal, 1(1), 33–45, 1992). Problems with the reliability of scores produced by such instruments have led to difficulties in interpreting the results of applied research (Schriesheim et al., Journal of Management, 19, 385–417, 1993). This paper has three aims. The first is to present a rationale for including information about the measurement properties of research instruments used in published accounting education research. Information regarding the reliability and validity of item scores is useful to readers, journal editors and reviewers - especially when the research topic addresses changes in educational policy and processes. Second, problems encountered with questionnairebased research are presented and current practice in accounting education is reviewed. Third, construct validation is discussed in respect of measurement validity and factor analytic methods. The paper concludes with suggestions for improving contemporary practice in accounting education research.     

Exploring Comparative Ratings and Constituent Facets of Public Trust in Risk Regulatory Bodies and Related Stakeholder Groups

This paper reports on two comparative ranking tasks performed by a sample of the British citizens (N = 304). The first was designed to compare levels of relative trust vested in a sample of UK risk regulatory bodies and associated stakeholder groups. The second sought to elicit a ranking of a range of previously identified facets of social trust referenced to their desirability as attributes of a government funded risk regulatory body. The ranking tasks were embedded within a broader programme of research focused on “Evaluating public understandings of and trust in the Health and Safety Executive” (Pidgeon et al., 2003). It is argued that deriving rankings of multi‐faceted phenomena using the method of paired comparisons offers a more robust approach to rating social trust entities than the direct ranking techniques used in previous studies in this area. Results are discussed with reference to qualitative findings from the broader programme of work on public trust in HSE (Pidgeon et al., 2003) and the wider literature on public trust in risk regulation.     

Intended or Unintended Consequences? Resource Erosion in New Zealand's Government Departments

Internationally, a feature of public sector reforms over the last twenty years is the characterisation of negative effects as unintended consequences, yet the reform programme continues as before (Humphrey et al., 1993 and 1998). This article reports the findings from an investigation of ostensibly unintended consequences of New Zealand's public sector financial management system: resource erosion in government departments which both escapes parliamentary scrutiny and damages departments' capability to perform even core functions (State Services Commission, 1998a; and Controller and Auditor–General, 1999). The findings suggest that these effects might not be unintended. The system's structures and processes, including designed–in resource erosion processes and a surreptitious approach, are highly consistent with those advocated to facilitate privatisation.     

The Business Risk Audit Approach and Audit Production Efficiency

Essentially, this study asks: Does the business risk audit (BRA) approach increase audit production efficiency? To answer this question empirically, direct and indirect tests are employed using proprietary, working paper data from the larger clients of a major Australian public sector audit provider and an efficiency frontier analytic methodology, data envelopment analysis (DEA). Results based on this proprietary, audit hours data for audit engagements carried out just after BRA approach implementation show that they have high levels of production efficiency and are risk-adjusted, with no significant difference in production efficiency between higher and lower business risk audit engagements. Results based on audit fees data for audit engagements carried out shortly before and after BRA approach implementation show that overall production efficiency significantly improves. Importantly, while this improvement is significant for lower-risk audit engagements, there is no significant improvement for higher-risk audit engagements. In the context of this study's research site, this is consistent with the BRA approach addressing inefficiencies created when lower-risk audit engagements are being over-audited. That is, the BRA approach can result in both risk-adjusted and more efficiently produced audits. With the re-emergence of the BRA approach in the literature and in practice, this study provides empirical evidence to support the claim that this audit approach can lead to ‘creating auditing efficiencies’ (Bell et al., 1997, p. 1).     

A Comparison of State-of-the-Art Classification Techniques for Expert Automobile Insurance Claim Fraud Detection

Several state‐of‐the‐art binary classification techniques are experimentally evaluated in the context of expert automobile insurance claim fraud detection. The predictive power of logistic regression, C4.5 decision tree, k‐nearest neighbor, Bayesian learning multilayer perceptron neural network, least‐squares support vector machine, naive Bayes, and tree‐augmented naive Bayes classification is contrasted. For most of these algorithm types, we report on several operationalizations using alternative hyperparameter or design choices. We compare these in terms of mean percentage correctly classified (PCC) and mean area under the receiver operating characteristic (AUROC) curve using a stratified, blocked, ten‐fold cross‐validation experiment. We also contrast algorithm type performance visually by means of the convex hull of the receiver operating characteristic (ROC) curves associated with the alternative operationalizations per algorithm type. The study is based on a data set of 1,399 personal injury protection claims from 1993 accidents collected by the Automobile Insurers Bureau of Massachusetts. To stay as close to real‐life operating conditions as possible, we consider only predictors that are known relatively early in the life of a claim. Furthermore, based on the qualification of each available claim by both a verbal expert assessment of suspicion of fraud and a ten‐point‐scale expert suspicion score, we can compare classification for different target/class encoding schemes. Finally, we also investigate the added value of systematically collecting nonflag predictors for suspicion of fraud modeling purposes. From the observed results, we may state that: (1) independent of the target encoding scheme and the algorithm type, the inclusion of nonflag predictors allows us to significantly boost predictive performance; (2) for all the evaluated scenarios, the performance difference in terms of mean PCC and mean AUROC between many algorithm type operationalizations turns out to be rather small; visual comparison of the algorithm type ROC curve convex hulls also shows limited difference in performance over the range of operating conditions; (3) relatively simple and efficient techniques such as linear logistic regression and linear kernel least‐squares support vector machine classification show excellent overall predictive capabilities, and (smoothed) naive Bayes also performs well; and (4) the C4.5 decision tree operationalization results are rather disappointing; none of the tree operationalizations are capable of attaining mean AUROC performance in line with the best. Visual inspection of the evaluated scenarios reveals that the C4.5 algorithm type ROC curve convex hull is often dominated in large part by most of the other algorithm type hulls.     

Resistance (to Fraud) Is Futile

This article studies a static principal–agent model of insurance fraud using a costly state verification approach. In an economy where there are two types of agents, the Truths, who always report the true state of the world, and the Dares, who dare misreport the true state of the world, I show that no separating contract exists. Furthermore, if the proportion of Dares is large enough, then the pooling contract, the amount of fraud and the number of agents found to have committed fraud are independent of the Dares' exact proportion in the economy. Finally, I show that investment in prevention can be useless if the proportion of Dares is large enough, which means that investing in prevention becomes a waste of resources. This last result holds when the proportion of Dares is large. When their proportion is small, investing in prevention reduces fraud.     

Pricing and hedging guaranteed minimum withdrawal benefits under a general Lévy framework using the COS method

This paper extends the Fourier-cosine (COS) method to the pricing and hedging of variable annuities embedded with guaranteed minimum withdrawal benefit (GMWB) riders. The COS method facilitates efficient computation of prices and hedge ratios of the GMWB riders when the underlying fund dynamics evolve under the influence of the general class of Lévy processes. Formulae are derived to value the contract at each withdrawal date using a backward recursive dynamic programming algorithm. Numerical comparisons are performed with results presented in Bacinello et al. [Scand. Actuar. J., 2014, 1–20], and Luo and Shevchenko [Int. J. Financ. Eng., 2014, 2, 1–24], to confirm the accuracy of the method. The efficiency of the proposed method is assessed by making comparisons with the approach presented in Bacinello et al. [op. cit.]. We find that the COS method presents highly accurate results with notably fast computational times. The valuation framework forms the basis for GMWB hedging. A local risk minimisation approach to hedging intra-withdrawal date risks is developed. A variety of risk measures are considered for minimisation in the general Lévy framework. While the second moment and variance have been considered in existing literature, we show that the Value-at-Risk (VaR) may also be of interest as a risk measure to minimise risk in variable annuities portfolios.     

Company takeovers and the auditor choice decision

There is a growing literature on how companies select their auditors. This literature emphasizes both theoretical development and empirical testing. Anderson, Stokes, and Zimmer (1993) recently identified corporate takeovers as events that generate auditor selection decisions; the acquiring company (the acquirer) has to choose whether to retain the new subsidiary’s (the acquiree) incumbent auditor or switch to its own auditor. To date, this is the only reported study that examines auditor changes following corporate acquisitions. Anderson et al. (1993) speculated that if the business activities of the acquirer and the acquiree are different, then the acquiree’s incumbent auditor is more likely to be retained. They used takeover type (horizontal, vertical, conglomerate) as a proxy for differences in activities. Their empirical results, however, were contrary to these expectations.The purpose of this paper is to test the robustness of Anderson et al.’s findings in a different national setting (Britain instead of Australia). Further, it modifies Anderson et al.’s test procedures by introducing new variables that are also hypothesized to affect auditor choice in acquired companies. The results show that proxies for differences in activities between acquirers and acquirees, namely takeover type and industry, are associated with the retention of the acquiree’s incumbent auditor. The takeover type finding therefore differs from the findings of Anderson et al. (1993).Other factors associated with the decision to retain auditors include audit opinions, contested takeover bids, changes in top management, restructuring and integration of businesses, and relative audit fees. Some of these factors have also been identified in other, non-takeover contexts. Finally, audit switches to Big Eight firms are found to be more likely than switches away from the Big Eight.