Accounting Fraud Detection Research Articles

Accounting fraud detection using contextual language learning

Accounting fraud is a widespread problem that causes significant damage in the economic market. Detection and investigation of fraudulent firms require a large amount of time, money, and effort for corporate monitors and regulators. In this study, we explore how textual contents from financial reports help in detecting accounting fraud. Pre-trained contextual language learning models, such as BERT, have significantly advanced natural language processing in recent years. We fine-tune the BERT model on Management Discussion and Analysis (MD&A) sections of annual 10-K reports from the Securities and Exchange Commission (SEC) database. Our final model outperforms the textual benchmark model and the quantitative benchmark model from the previous literature by 15% and 12%, respectively. Further, our model identifies five times more fraudulent firm-year observations than the textual benchmark by investigating the same number of firms, and three times more than the quantitative benchmark. Optimizing this investigation process, where more fraudulent observations are detected in the same size of the investigation sample, would be of great economic significance for regulators, investors, financial analysts, and auditors.

Detecting accounting fraud in family firms: Evidence from machine learning approaches

The primary objective of this research is to detect accounting fraud in Chinese family firms through the utilization of imbalanced ensemble learning algorithms. It serves as the first endeavor to predict fraud in family firms using machine learning algorithms, thus addressing the gap in machine-learning modeling for family business research. The findings of this study demonstrate that the ensemble learning models exhibit superior effectiveness in identifying accounting fraud compared to the logistic regression approach. Moreover, the imbalanced ensemble learning classifiers outperform the conventional models. Significantly, among all the studied fraud classifiers, the CUSBoost classifier consistently attains the best overall performance. This research contributes to the field of accounting fraud detection in family firms by shifting the focus from conventional causal inference methods (such as regression) to machine-learning-based predictive techniques. Additionally, it extends existing literature on accounting fraud detection by emphasizing the issue of data imbalance in fraud datasets and demonstrating the superiority of imbalanced machine learning algorithms over conventional approaches in detecting accounting fraud.

Disentangling the black box around CEO and financial information-based accounting fraud detection: machine learning-based evidence from publicly listed U.S. firms

AbstractThis study investigates the predictive power of CEO characteristics on accounting fraud utilizing a machine learning approach. Grounded in upper echelons theory, we show the predictive value of widely neglected CEO characteristics for machine learning-based accounting fraud detection in isolation and as part of a novel combination with raw financial data items. We employ five machine learning models well-established in the accounting fraud literature. Diverging from prior studies, we introduce novel model-agnostic techniques to the accounting fraud literature, opening further the black box around the predictive power of individual accounting fraud predictors. Specifically, we assess CEO predictors concerning their feature importance, functional association, marginal predictive power, and feature interactions. We find the isolated CEO and combined CEO and financial data models to outperform a no-skill benchmark and isolated approaches by large margins. Nonlinear models such as Random Forest and Extreme Gradient Boosting predominantly outperform linear ones, suggesting a more complex relationship between CEO characteristics, financial data, and accounting fraud. Further, we find CEO Network Size and CEO Age to contribute second and third strongest towards the best model’s predictive power, closely followed by CEO Duality. Our results indicate U-shaped, L-shaped, and weak L-shaped associations for CEO Age, CEO Network Size, CEO Tenure, and accounting fraud, consistent with our superior nonlinear models. Lastly, our empirical evidence suggests that older CEOs who are not simultaneously serving as chairman and CEOs with an extensive network and high inventory are more likely to be associated with accounting fraud.

Predicting accounting fraud using imbalanced ensemble learning classifiers – evidence from China

AbstractThe current research aims to launch effective accounting fraud detection models using imbalanced ensemble learning algorithms for China A‐Share listed firms. Based on a sample of 33,544 Chinese firm‐year instances from 1998 to 2017, this research respectively established one logistic regression and four ensemble learning classifiers (AdaBoost, XGBoost, CUSBoost, and RUSBoost) by 12 financial ratios and 28 raw financial data. Additionally, we divided the sample into the train and test observations to evaluate the classifiers' out‐of‐sample performance. In detail, we applied two metrics, namely, Area under the ROC (receiver operating characteristic) curve (AUC) and Area under the Precision‐Recall curve (AUPR), to evaluate classifiers' discriminability. In the supplement test, this study put forward an algebraic fused model on the basis of the four ensemble learning classifiers and introduced the sliding window technique. The empirical results showed that the ensemble learning classifiers can detect accounting fraud for the imbalanced China A‐listed firms far more effectively than the logistic regression model. Moreover, imbalanced ensemble learning classifiers (CUSBoost and RUSBoost) effectively performed better than the common ensemble learning models (AdaBoost and XGBoost) in average. The algebraic fused model in the supplement test also obtained the highest average AUC and AUPR among all the employed algorithms. Our results offer firm support for the potential role of Machine Learning (ML)‐based Artificial Intelligence (AI) approaches in reliably predicting accounting fraud with high accuracy. Similarly, for the Chinese settings, our ML‐based AI offers utmost advantage in forecasting accounting fraud. Finally, this paper fills the research gap on the applications of imbalanced ensemble learning in accounting fraud detection for Chinese listed firms.

Bibliometric Analysis of Accounting Fraud and Earnings Management During the period 2012-2022

This study performs a bibliometric analysis of articles on accounting fraud and earnings management published in the period from 2012 to 2022. The bibliometric analysis makes it possible to evaluate the production and impact of the scientific literature on the subject under study. To carry out the analysis, the Scopus database was used to search for papers containing the terms “accounting fraud” and/or “earnings management”. A total of 688 articles were analyzed. The results show that the United Kingdom, the USA and several Asian countries have published more in this field. Likewise, the journals that have published the most articles on this subject are the Journal of Applied, Business Research, Journal of Business Research and Corporate Ownership and Control. And, the main topics of the articles deal with the relationship between earnings management and corporate governance, ethical aspects, auditing and regulation, together with accounting fraud detection techniques. The study shows a great interest in this subject in the scientific community.

Detecting accounting fraud in companies reporting under US GAAP through data mining

The accounting fraud detection models developed on financial data prepared under US Generally Accepted Accounting Principles (GAAP) in the current literature achieve significantly weaker performance than models based on financial data prepared under different accounting standards. This study contributes to the US GAAP accounting fraud data mining literature through the attainment of higher model performance than that reported in the prior literature. Financial data from the 10-K forms of 320 fraudulent financial statements (80 fraudulent companies) and 1,200 nonfraudulent financial statements (240 nonfraudulent companies) were collected from the US Security and Exchange Commission. The eight most commonly used data mining techniques were applied to develop prediction models. The results were cross-validated on a testing dataset and then compared according to parameters of accuracy, F-measure, and type I and II errors with existing studies from the US, China, Greece, and Taiwan. As a result, the developed predictive models for accounting fraud achieved performance comparable to those achieved by models built on data from other accounting standards. Moreover, the developed models also significantly outperformed (accuracy 10.5%, F-measure 16.1%, type I error 12.2% and type II error 15.2%) existing studies based on US GAAP financial data. Furthermore, this study provides an extensive literature review encompassing recent accounting fraud theory. It enhances the existing US fraud data mining literature with a performance comparison of studies based on other accounting standards.

A novel integrated AHP and MULTIMOORA method with interval-valued spherical fuzzy sets and single-valued spherical fuzzy sets to prioritize financial ratios for financial accounting fraud detection

Many studies have used different financial ratios for financial accounting fraud detection. This study focuses on multicriteria decision-making (MCDM) for ranking 25 financial ratios with respect to six criteria in detecting financial accounting fraud using interval-valued spherical fuzzy sets (IVSFS) and single-valued spherical fuzzy sets (SVSFS) to overcome uncertainties in decision-making process of financial analysts. This study proposes an integrated Analytic Hierarchy Process (AHP) and Multi-Objective Optimization by a Ratio Analysis plus the Full Multiplicative Form (MULTIMOORA) approach using IVSFS and SVSFS. Comparative results are obtained and discussed in prioritization of financial ratios for both IVSFS and SVSFS.

METRICS IN SMALL-SIZED QURAN DATASET FOR BENFORD’S LAW

Benford’s law is widely applied in testing anomalies in various dataset, including accounting fraud detection and population numbers. It is a statistical regularity, which is said that it works better with larger datasets that span large orders of magnitude distributed in a non-uniform way. In this study, we examine the potential metrics in small-sized Quran dataset that are applicable for the Benford’s law. Against our expectations, we find that the Quran dataset conforms to the Benford’s law. We provide evidence that metrics such as total paragraph per chapter and total verse per chapter conform to Benford’s distribution. However, total verse is closer to Benford’s law prediction compared to total paragraph.

THE USE OF MACHINE LEARNING TECHNIQUES IN ACCOUNTING. A SHORT SURVEY

Machine learning (ML) is a subset of artificial Intelligence (AI) aiming to develop systems that can learn and continuously improve the abilities through generalization in an autonomous manner. ML is presently all around us, almost every facet of our digital and real life is embedding some ML related content. Customer recommendation systems, customer behavior prediction, fraud detection, speech recognition, image recognition, black & white movies colorization, accounting fraud detection are just some examples of the vast range of applications in which ML is involved. The techniques that this paper investigates are mainly focused on the use of neural networks in accounting and finance research fields. An artificial neural network is modelling the brain ability of learning intricate patterns from the information presented at its inputs using elementary interconnected units, named neurons, grouped in layers and trained by means of a learning algorithm. The performance of the network depends on many factors like the number of layers, the number of each neurons in each layer, the learning algorithm, activation functions, to name just a few of them. Machine learning algorithms have already started to replace humans in jobs that require document’s processing and decision making.

Use of Benford's law on academic publishing networks

Benford's law, also known as the first-digit law, has been widely used to test for anomalies in various data ranging from accounting fraud detection, stock prices, and house prices to electricity bills, population numbers, and death rates. Scientific collaboration graphs have been studied extensively as data availability increased. Most research was oriented towards analysing patterns and typologies of citation graphs and co-authorship graphs. Most countries group publications into categories in an attempt to objectively measure research output. However, the scientific community is complex and heterogeneous. Additionally, scientific fields may have different publishing cultures, which make creating a unified metric for evaluating research output problematic. In complex systems like these, it is important to regularly observe potential anomalies and examine them more carefully in an attempt to either improve the evaluation model or find potential loopholes and misuses. In this paper, we examine the potential application of Benford's law on the official research database of Slovenia. We provide evidence that metrics such as number of papers per researcher conform to Benford's distribution, while the number of authors per paper does not. Additionally, we observe some anomalies and provide potential reasoning behind them.

On the possibility of Benfordʼs Law application in accounting fraud detection

Purpose: The purpose of the article is to indicate that, theoretically and practically, Benford’s Law can be applied in order to detect accounting frauds. Methodology/approach: The paper provides an overview of current regulations and experts’ opinions published in the existing literature and internet sources. Moreover, data analysis was used as a research method. Findings: The results of assessing the conformity of the first two significant digits of distribution of foreign revenues from the sales of finished products to Benford’s Law (using the chi-square goodness of fit test) showed that in the case of a proper accounting process, one cannot reject the hypothesis that the data conform to Benford’s Law. On the other hand, the analysis of intentionally falsified foreign revenues led to the conclusion that in the case of an improper accounting process, data, in general, does not conform to Benford’s Law. Research limitations/implications: In the study, it was assumed that the human mind generates false val-ues of accounting entries, and the number of attempts to commit fraud was limited to 10. Originality/value: The article extends the existing knowledge of using Benfordʼs Law in detecting ac-counting fraud in the Polish literature.

부정적 사건 전 내부자거래 행태

We investigate the insider trading behavior before the disclosure of negative events. Specifically, we examine whether insider sales increase before a negative event occurs, when they concentrate, and whether corporate governance plays an active role in monitoring insider trading.BRThe major findings of the study are as follows.BRFirst, the stock prices of firms with negative events decline about 10 months before the negative events. Insider trades by executives and largest shareholders show significant net selling behavior at least 24 months before the negative events. It is interpreted that insiders with private information about firm’s future negative events sell stock in advance to avoid losses. Second, foreign investors or institutional investors do not play a significant role in monitoring opportunistic insider trading before negative events such as accounting fraud detection and administrative issues designation. Third, insider sales of ‘10% or more major shareholders’ occurred much later than those of ‘executives and largest shareholders’. It could be due to the different ability of the insider to access the inside information of the firm.BRThis study extends the scope of related research by analyzing the insider transactions that occur before negative events such as detection of fraud or administrative issues designation in Korea. In addition, the results of this study show that insider trading occurs two years before the negative event, which provides useful insights for the regulatory authorities in charge of preventing unfair insider trading in the capital market.

Auditor Tenure and Accounting Fraud Detection

Auditor Tenure and Accounting Fraud Detection

Artificial Intelligence and Fraud Detection

Fraud exists in all walks of life and detecting and preventing fraud represents an important research question relevant to many stakeholders in society. With the rise of big data and artificial intelligence, new opportunities have arisen in using advanced machine learning models to detect fraud. This chapter provides a comprehensive overview of the challenges in detecting fraud using machine learning. We use a framework (data, method, and evaluation criterion) to review some of the practical considerations that may affect the implementation of ma-chine-learning models to predict fraud. Then, we review select papers in the academic literature across different disciplines that can help address some of the fraud detection challenges. Finally, we suggest promising future directions for this line of research. As accounting fraud constitutes an important class of fraud, we will discuss all of these issues within the context of accounting fraud detection.

DETECTION MODELS FOR UNINTENTIONAL FINANCIAL RESTATEMENTS

The aim of manuscript is to analyze and identify determinants of honest accounting errors leading to financial restatements based on data from SEC database and from annual reports. Reason for this study is that accounting errors are expensive for companies that need to change already published financial statements and have impact on company reputation and stock price. Most of authors focus on prediction of accounting frauds and financial restatements remain in the background of research. This study initially tests existing accounting fraud detection model of Beneish on a sample of 40 financial restatement companies over 10 years and develops two new pioneer prediction models, one based on linear discriminant analysis (LDA) and another based on logistic regression. In testing dataset, LDA model has achieved accuracy 70.96%, specificity 25.00% and sensitivity 79.83% and logistic regression model has achieved accuracy 62.22%, specificity 41.66% and sensitivity 66.67%, performance of both models is better than existing Beneish model or other studies in this field. Developed models can be widely used by both internal and external users of financial statements, who would like to determine if financial statements of analyzed company include accounting errors or not, thanks to easily interpretable results in equation form.

Credit Rating Agencies and Accounting Fraud Detection

This study examines whether and when credit rating agencies take negative rating actions against issuers that commit accounting fraud before the fraud is publicly revealed and investigates the economic effects of such rating actions. After controlling for firms’ economic fundamentals and other sources of information, we find that Standard & Poor’s (S&P), an issuer-paid rating agency, takes negative rating actions against fraud firms, including downgrades and negative credit watches, as early as four quarters before the revelation of fraud. Egan-Jones, an investor-paid rating agency with no access to management and material nonpublic information, is slower than S&P to downgrade fraud firms, especially when information uncertainty is high and when fraud firms are subject to private SEC investigations. Last, we find that S&P’s negative rating actions against fraud firms are informative to the market and are associated with a shorter fraud duration. Taken together, our results suggest that credit rating agencies use their information advantage to detect accounting fraud.

Rasio-Rasio Keuangan dalam Memprediksi Kecurangan

Penlitian ini bertujuan untuk membuktikan apakah perusahaan yang melakukan kecurangan akuntansi dengan yang tidak melakukan kecurangan akuntansi memiliki nilai rasio yang berbeda secara signifikan dengan menggunakan dua model pendeteksi kecurangan akuntansi yaitu Beneish M-Score da Altman Z-Score. Penelitian ini menggunakan delapan variabel yaitu debt to equity ratio , debt to total asset ratio, net profit to revenue ratio, current asset to total asset ratio, receivable to revenue ratio, inventory to total asset ratio,working capital to total asset ratio dan revenue to total asset ratio.
 Populasi dari penelitian ini adalah perusahaan manufaktur yang terdaftar di Bursa Efek Indonesia periode 2010-2016. Total sampel dalam penelitian ini sebanyak 57 perusahaan dengan jumlah observasi sebanyak 212. Pengujian hipotesis dengan menggunakan uji beda independent sample t-test.
 Hasil penelitian ini menunjukkan bahwa variabel debt to equity ratio , debt to total asset ratio, net profit to revenue ratio, working capital to total asset ratio dan revenue to total asset ratio memiliki perbedaan yang signifikan antara perusahaan yang melakukan kecurangan akuntansi dengan yang tidak melakukan kecurangan akuntansi. Sedangkan untuk variabel current asset to total asset ratio, receivable to revenue ratio , dan inventory to total asset ratio tidak memiliki perbedaan yang signifikan antara perusahaan yang melakukan kecurangan akuntansi dengan yang tidak melakukan kecurangan akuntansi.

Researches of Detection of Fraudulent Financial Statements Based on Data Mining

Financial statement fraud has been one of the biggest challenges in the modern business world. Financial accounting fraud detection (FAFD) has become an emerging topic of great importance for academic, research and industries. In this paper, the effectiveness of Data Mining (DM) classification techniques in detecting firms that issue fraudulent financial statements (FFS) and deals with the identification of factors associated to FFS are explored. Our study investigates the usefulness of Data Mining techniques including Decision Trees, Neural Networks and Bayesian Belief Networks in the identification of fraudulent financial statements. At last, we compare the three models in terms of their performances.

Responsibility of an auditor for accounting fraud detection

Consequences of detected accounting frauds are shared by all stakeholders starting from the enterprise, through investors, creditors, business partners, the state, to the audit firm. Responsibility for the true and fair presentation of the financial statements lies with managers, auditors and regulators. The aim of the paper is to determine the level of responsibility of auditors in detecting any manipulative acts and fraud. In this paper the authors will also reflect on the conduct of auditors encouraged by personal or commercial pressure which leads to disruption of audit quality and public confidence in the audit report.

결제로그 분석 및 데이터 마이닝을 이용한 이상거래 탐지 연구 조사

Due to a rapid advancement in the electronic commerce technology, the payment method varies from cash to electronic settlement such as credit card, mobile payment and mobile application card. Therefore, financial fraud is increasing notably for a purpose of personal gain. In response, financial companies are building the FDS (Fraud Detection System) to protect consumers from fraudulent transactions. The one of the goals of FDS is identifying the fraudulent transaction with high accuracy by analyzing transaction data and personal information in real-time. Data mining techniques are providing great aid in financial accounting fraud detection, so it have been applied most extensively to provide primary solutions to the problems. In this paper, we try to provide an overview of the research on data mining based fraud detection. Also, we classify researches under few criteria such as data set, data mining algorithm and viewpoint of research.