In the rise of digital transformation and big data, accounting and auditing professions are required to adopt advanced techniques that help detect irregularities and frauds. This chapter collects 1,112 documents from Scopus database published in 288 sources over the period 2008-2021. A bibliometric analysis is used to depict trends, and applicable methodologies are adopted to build a comprehensible base that will serve to instill a new fraud detection model. More specifically, RStudio through “biblioshiny package”, VOSviewer, and Excel are the tools embraced to analyze the dataset information. The chapter explores the literature growth over time and addresses key aspects of the literature, such as most relevant documents, authors, countries, citation analysis. This chapter applies a network and content analysis using the bibliographical coupling, trend analysis, word cloud, and co-occurrence analysis. The theoretical model helps auditors, forensic accountants, top managements, analysts, and policy makers predict potential anomalies and misstatements.
TopIntroduction
The Association of Certified Fraud Examiners (ACFE) recently released the “Report to the Nations: 2020 Global study on Occupational Fraud and Abuse” which covers 2,504 fraud cases from 125 countries that caused total loss of more than $ 3.6 billion. Organizations worldwide lose about five percent of their annual revenue to fraud with a median loss of $125,000 per case and an average loss of $1,509,000 per case (ACFE, 2020). The report concludes that the corruption is the most common scheme in every global region. On the other hand, 86% of fraud cases are due to the misappropriation of assets which appears to be the least costly ($100,000 median loss) while 10% are due to financial statement manipulation that appears to be the costliest ($954,000 median loss).
In parallel, a rising trend highlights the emergence of forensic accounting that includes a wide range of activities such as litigation support, expert witness, and fraud investigation. Forensic accounting is “the science of gathering and presenting financial information in a form that will be accepted by a court of jurisprudence against perpetrators of economic crimes” (Manning & CFE, 1999). According to the American Institute of Certified Public Accountants (AICPA), “Forensic accounting is the application of accounting principles, theories and discipline to facts or hypotheses at issues in a legal dispute and encompasses every branch of accounting knowledge.”
In the present study, the authors collect 1,112 previous publications related to the said topic from Scopus Database. The authors conduct the analysis using RStudio, VOSviewer, and Excel. The analysis will serve to derive important conclusions such as: similarities in fraud circumstances, main reasons, methodologies to depict and prevent frauds, technological advances in fraud detection, and qualitative variables such as CEO behavioral patterns, ethics, gender, and compensation biases along with whistleblowing incentives and employees’ training programs that aim at signaling early fraud stages.
Deep data analysis, clustering methods, data mining, artificial intelligence and expert system application are expected to shift the knowledge base. Some common techniques such as Beneish Model, Benford’s Law and other sampling techniques are used in the forensic accounting investigations. Foremost, trend analysis, horizontal and vertical analysis, regression analysis, ratio analysis and Markov chain are applied. Computer-based programs are managed to analyze quantitative and qualitative data that have a significant association with financial information manipulation. But what is the optimal model to depict a fraud? Can forensic auditors generalize a unique model? In fact, such models remain inconsistent and dissatisfactory randomized. Concomitantly, some techniques suffer from major drawbacks. To tackle the problem, one should address the topic from an aggregate perspective where auditors and researchers gather data from fraudulent companies, perform analytical procedures and approach their similarities based on clustering techniques that account for country-industry-firm levels. The objective is to generate a new model that combines classical techniques with digital ones which will help depict anomalies and provide valuable insights to deter potential frauds.
The authors contribute to the literature by attempting to introduce a new model based on previous literature in the studied field. It addresses a basic and essential question: What major role would be played by forensic accountants? Are they able to reduce fraud attempts? How would the analysis of the previous literature lead to find the best model to predict and early detect fraud?
In fact, to detect anomalies, most of the previous discern the different facets of the audit process. This study conducts a bibliometric analysis that constitutes a framework to generalize results and methods. It aims at contributing to the knowledge of auditors and forensic accountants. It provides stakeholders with a window of credible information free from earnings manipulations and sheds light on the quality of firms’ financial reporting. The study is novel as it suggests recommendations to further test the model and gradually refine it to suit all case scenarios.
The remainder of the chapter is structured as follows: Section 2 discusses the background with relevant literature review. Section 3 describes the research methodology, while Section 4 presents the bibliometric findings and presents a proposed fraud detection model. Section 5 concludes, while section 6 suggests solutions, recommendations, and future research directions.