Purpose:This study aims to integrate large language models(LLMs)with interpretable machine learning methods to develop a multimodal data-driven framework for predicting corporate financial fraud,addressing the limitat...Purpose:This study aims to integrate large language models(LLMs)with interpretable machine learning methods to develop a multimodal data-driven framework for predicting corporate financial fraud,addressing the limitations of traditional approaches in long-text semantic parsing,model interpretability,and multisource data fusion,thereby providing regulatory agencies with intelligent auditing tools.Design/methodology/approach:Analyzing 5,304 Chinese listed firms’annual reports(2015-2020)from the CSMAD database,this study leverages the Doubao LLMs to generate chunked summaries and 256-dimensional semantic vectors,developing textual semantic features.It integrates 19 financial indicators,11 governance metrics,and linguistic characteristics(tone,readability)with fraud prediction models optimized through a group of Gradient Boosted Decision Tree(GBDT)algorithms.SHAP value analysis in the final model reveals the risk transmission mechanism by quantifying the marginal impacts of financial,governance,and textual features on fraud likelihood.Findings:The study found that LLMs effectively distill lengthy annual reports into semantic summaries,while GBDT algorithms(AUC>0.850)outperform the traditional Logistic Regression model in fraud detection.Multimodal fusion improved performance by 7.4%,with financial,governance,and textual features providing complementary signals.SHAP analysis revealed financial distress,governance conflicts,and narrative patterns(e.g.,tone anchoring,semantic thresholds)as key fraud indicators,highlighting managerial intent in report language.Research limitations:This study identifies three key limitations:1)lack of interpretability for semantic features,2)absence of granular fraud-type differentiation,and 3)unexplored comparative validation with other deep learning methods.Future research will address these gaps to enhance fraud detection precision and model transparency.Practical implications:The developed semantic-enhanced evaluation model provides a quantitative tool for assessing listed companies’information disclosure quality and enables practical implementation through its derivative real-time monitoring system.This advancement significantly strengthens capital market risk early warning capabilities,offering actionable insights for securities regulation.Originality/value:This study presents three key innovations:1)A novel“chunking-summarizationembedding”framework for efficient semantic compression of lengthy annual reports(30,000 words);2)Demonstration of LLMs’superior performance in financial text analysis,outperforming traditional methods by 19.3%;3)A novel“language-psychology-behavior”triad model for analyzing managerial fraud motives.展开更多
Financial fraud arises from the exaggeration of business interests,and an accurate detection or prediction is a useful tool for both corporate management and capital market systems.A collection of computer technologie...Financial fraud arises from the exaggeration of business interests,and an accurate detection or prediction is a useful tool for both corporate management and capital market systems.A collection of computer technologies has been made on this problem so far,and one of the most important solutions is unsupervised learning algorithms.Among them,most approaches work by analysing the internal relations in financial data and finding a new description of non-fraud firms.However,current studies focus a lot on the geometry attribute of financial data,while overlooking the obvious behaviour patterns and peer effects among firms.This has limited the accuracy of representation and furthermore the detection performance.In this work,a very general class of functions is allowed to represent firms,constraining them by peer effects between firms and presenting an error-distribution-based financial fraud firm detection approach.Experimental results have shown great performance of the proposed approach.展开更多
Taking the financial fraud of RX Company in 2020 as a case, this paper collects and studies the investigation report of RX Company, studies the financial fraud means of RX Company from the perspective of investor prot...Taking the financial fraud of RX Company in 2020 as a case, this paper collects and studies the investigation report of RX Company, studies the financial fraud means of RX Company from the perspective of investor protection, and analyzes the fraud motivation of RX Company in detail from the perspectives of internal and external causes. Finally, this event triggered thinking on how to implement feasible measures to protect the interests of investors.展开更多
As financial criminal methods become increasingly sophisticated, traditional anti-money laundering and fraud detection approaches face significant challenges. This study focuses on the application technologies and cha...As financial criminal methods become increasingly sophisticated, traditional anti-money laundering and fraud detection approaches face significant challenges. This study focuses on the application technologies and challenges of big data analytics in anti-money laundering and financial fraud detection. The research begins by outlining the evolutionary trends of financial crimes and highlighting the new characteristics of the big data era. Subsequently, it systematically analyzes the application of big data analytics technologies in this field, including machine learning, network analysis, and real-time stream processing. Through case studies, the research demonstrates how these technologies enhance the accuracy and efficiency of anomalous transaction detection. However, the study also identifies challenges faced by big data analytics, such as data quality issues, algorithmic bias, and privacy protection concerns. To address these challenges, the research proposes solutions from both technological and managerial perspectives, including the application of privacy-preserving technologies like federated learning. Finally, the study discusses the development prospects of Regulatory Technology (RegTech), emphasizing the importance of synergy between technological innovation and regulatory policies. This research provides guidance for financial institutions and regulatory bodies in optimizing their anti-money laundering and fraud detection strategies.展开更多
Luckin Coffee grew rapidly in the past few years and it was the fastest Chinese Concept company go public in NASDAQ.However,its stock price plunged in 2020 due to financial fraud.It is crucial for Luckin Coffee to thi...Luckin Coffee grew rapidly in the past few years and it was the fastest Chinese Concept company go public in NASDAQ.However,its stock price plunged in 2020 due to financial fraud.It is crucial for Luckin Coffee to think about how to recover from the financial fraud and become a business representative of the“new retail”under the Internet model again.To recover from the financial crisis caused and gain profit,Luckin could improve its existing business model by improving product development,strengthening capital management in both marketing and expansion.The internal structure can also be adjusted by diversifying equity and replacing the manager team.Taking Luckin Coffee as an example,this paper analyzes the causes of financial fraud,propose solutions to recover from it,and evaluate the effectiveness of solutions in the context of reality.展开更多
Financial fraud,which has become a global issue,is a subject of discussion,surpassing time.Financial fraud significantly undermines investors’confidence and affects the health of capital markets.Hence,it is valuable ...Financial fraud,which has become a global issue,is a subject of discussion,surpassing time.Financial fraud significantly undermines investors’confidence and affects the health of capital markets.Hence,it is valuable to explore the reasons for committing financial fraud and propose solutions to this issue.This paper focuses on two financial fraud cases in recent years,Toshiba in 2015 and Luckin Coffee in 2020,analyzes and compares the reasons for the financial fraud in terms of pressure and opportunity factors,as well as proposes comprehensive suggestions for dealing with the corporate financial fraud issue.展开更多
The landscape of financial transactions has grown increasingly complex due to the expansion of global economic integration and advancements in information technology.This complexity poses greater challenges in detecti...The landscape of financial transactions has grown increasingly complex due to the expansion of global economic integration and advancements in information technology.This complexity poses greater challenges in detecting and managing financial fraud.This review explores the role of Graph Neural Networks(GNNs)in addressing these challenges by proposing a unified framework that categorizes existing GNN methodologies applied to financial fraud detection.Specifically,by examining a series of detailed research questions,this review delves into the suitability of GNNs for financial fraud detection,their deployment in real-world scenarios,and the design considerations that enhance their effectiveness.This review reveals that GNNs are exceptionally adept at capturing complex relational patterns and dynamics within financial networks,significantly outperforming traditional fraud detection methods.Unlike previous surveys that often overlook the specific potentials of GNNs or address them only superficially,our review provides a comprehensive,structured analysis,distinctly focusing on the multifaceted applications and deployments of GNNs in financial fraud detection.This review not only highlights the potential of GNNs to improve fraud detection mechanisms but also identifies current gaps and outlines future research directions to enhance their deployment in financial systems.Through a structured review of over 100 studies,this review paper contributes to the understanding of GNN applications in financial fraud detection,offering insights into their adaptability and potential integration strategies.展开更多
Fraud detection is an important task of financial supervision and financial risk management.Several fraud behaviors can be connected with cycle transactions in which the money initially sent from one bank account even...Fraud detection is an important task of financial supervision and financial risk management.Several fraud behaviors can be connected with cycle transactions in which the money initially sent from one bank account eventually returns back to the same account.In this paper,we propose an efficient method for detecting such kind of frauds from large-scale financial transaction data.The method first constructs a transaction graph after pre-processing the original data,then divides the graph into its strongly connected components,and finally uses multiple threads to enumerate temporal cycles on different components in a parallel manner.Existing temporal cycle enumeration algorithms usually constraint the length of the cycle or the size of the time-window,which are not suitable for the specific application of financial fraud detection.In light of this,we extend the classical Johnson algorithm to enumerate temporal cycles without length and time-window constraints.To further improve the efficiency of enumeration,we introduce a block-time mechanism that avoids unnecessary multiple explorations of the same parts of the graph components.Experiments show that our method,with multithreading,is on average 15–20 times,and even 100 times faster than the existing competitor.Additionally,we adopt strategies such as amount constraint during cycle enumeration,which assist in reducing the false-positive rate of detected frauds.展开更多
Large-scale accounting scandals which were reflected to the world public opinion particularly in the 2000s （such as Enron, Lucent, Xerox, and Parmalat Bank for Reconstruction） carried the matter of fraudulent financ...Large-scale accounting scandals which were reflected to the world public opinion particularly in the 2000s （such as Enron, Lucent, Xerox, and Parmalat Bank for Reconstruction） carried the matter of fraudulent financial reporting which was made to deceive the financial statement users （Fraudulent Financial Report （FFR）） to the forefront in the agenda of the academicians, operators, and regulatory authorities. As in every crime action, the most effective measure to be taken in preventing FFR events is to try to prevent the FFR before arising. In order to achieve this, in the most effective manner, FFR events should be determined in the formation process. In this study, fraudulent financial statements are tried to be determined by using financial ratios. For this, financial statements of 22 companies which transact in the textile industry in Istanbul Stock Exchange （ISE） were examined. Twenty-three financial ratios were selected for the purpose of determining the risk of fraudulence in the financial statements of the selected companies. These ratios increased in value by multiple regression analysis. The findings which were obtained in the study indicated that some financial statements had the risk of fraudulence. It was concluded that the ratios of inventory/current asset, total debt ratio, and equity turnover rate were a good indicator in the determination of fraudulent financial statements.展开更多
By summarizing the factor of the financial statement fraud in existing research outcome, the paper confirms the discriminating characteristic of the financial statement fraud and sets up a theoretic model to discrimin...By summarizing the factor of the financial statement fraud in existing research outcome, the paper confirms the discriminating characteristic of the financial statement fraud and sets up a theoretic model to discriminate the financial statement fraud. Using radial basis function neural network, regarding to the swatch that the listed company that is punished by the Securities Regulatory Commission or the Ministry of Finance, and according to the clustering elements, validating across by set one aside, the paper verifies respectively the 22 characteristics and 31 characteristics of discriminating model. According to the clustering elements, validating across by set one aside, the paper verifies respectively the 31 characteristics and 8 characteristics selected by Fisher-ratio of discriminating model. The research outcome indicates the discriminating ability of the model including 8 characteristics is better elevated than the traditional model including 31 characteristics by comparing the disciplinary error and the forecast precision.展开更多
This paper studied the ＂pressure＂ and ＂opportunity＂ factors that caused financial statement fraud on the basis of the data of 41 A-share listed companies （1999-2004） in China which had been forfeited by China Se...This paper studied the ＂pressure＂ and ＂opportunity＂ factors that caused financial statement fraud on the basis of the data of 41 A-share listed companies （1999-2004） in China which had been forfeited by China Securities Regulatory Commission （SRC） because of accounting irregularities. The author found that avoiding ＂ST＂ and ＂PT＂ was the primary pressure, and the opportunities mainly came from the higher top 10 shareholders＇ ownership concentration, the lower proportion of independent directors, the fewer number of directorate meetings and shares owned by the directorate members, board chairman and CEO held by one person and the ineffective supervisor boards. We also found that the companies involved financial statement fraud had the lower first majority shareholder＇s share proportion and they changed CPA firm more frequently.展开更多
To evaluate the applicability of the M-score model in the Chinese capital market,this research observed 190 financial fraud samples punished by the China Securities Regulatory Commission(CSRC)in the years from 2014 to...To evaluate the applicability of the M-score model in the Chinese capital market,this research observed 190 financial fraud samples punished by the China Securities Regulatory Commission(CSRC)in the years from 2014 to 2018.The test results indicate that two types of errors are high,which means that the applicability of the M-score is unacceptable.Therefore,in this paper,a 9-index model is constructed by Wald's backward stepwise regression method,and the optimal threshold is set by the Beneish expected cost method(ECM).The accuracy of the modified M-score is significantly improved,especially the Type I error rate of is reduced from 70.37%to 19.75%.The receiver operating characteristic(ROC)curve test also proves the superior identification effect of the modified M-score applied in the Chinese market.Finally,variables such as current ratio,fixed asset index,and equity concentration in the modified model could represent the fraud characteristics of Chinese listed companies.展开更多
Understanding the root causes of fraud and learning about the most effective fraud prevention mechanisms are critical in reducing the incidence of financial fraud. Therefore, this study solicits the views of fraud inv...Understanding the root causes of fraud and learning about the most effective fraud prevention mechanisms are critical in reducing the incidence of financial fraud. Therefore, this study solicits the views of fraud investigators on the existence and effectiveness of fraud prevention mechanisms within Government-Linked Companies （GLCs）. Specifically, effective fraud prevention mechanisms （FPM） should be undertaken by five critical groups, namely the board of directors, audit committees, external auditors, internal auditors, and anti-fraud specialists. A total of 147 questionnaires were distributed to fraud investigators of all public listed GLCs in Malaysia. Out of those, a total of 45 usable responses were received, which represents an effective response rate of 30.6%. In terms of existence, the respondents ranked ＂management review of internal controls＂ and ＂external audits of financial statements＂ as the top-most FPMs. This was followed by other mechanisms such as operational audits, internal audits and internal control review/improvements by departments. Out of the 27 fraud prevention mechanisms, fraud investigators perceiyed surprise audits, fraud hotline, fraud prevention program and training, anti-fraud policy, fraud vulnerability reviews, operational audits, whistle-blowing policy, internal audit or fraud examination department, and, imposing penalty and disciplinary action as more effective than the others. Overall, this study provides important insights to practitioners and organizations in identifying fraud prevention mechanisms that are most effective.展开更多
Different from foreign capital markets,china’s domestic capital markets are special,which also determines that the research on financial pressure starts from the reform of state-owned enterprises,and draws lessons fr...Different from foreign capital markets,china’s domestic capital markets are special,which also determines that the research on financial pressure starts from the reform of state-owned enterprises,and draws lessons from the relevant theories of financial risk and financial fraud,thus gradually forming a more diversified research results.展开更多
Information analysis of high dimensional data was carried out through similarity measure application. High dimensional data were considered as the a typical structure. Additionally, overlapped and non-overlapped data ...Information analysis of high dimensional data was carried out through similarity measure application. High dimensional data were considered as the a typical structure. Additionally, overlapped and non-overlapped data were introduced, and similarity measure analysis was also illustrated and compared with conventional similarity measure. As a result, overlapped data comparison was possible to present similarity with conventional similarity measure. Non-overlapped data similarity analysis provided the clue to solve the similarity of high dimensional data. Considering high dimensional data analysis was designed with consideration of neighborhoods information. Conservative and strict solutions were proposed. Proposed similarity measure was applied to express financial fraud among multi dimensional datasets. In illustrative example, financial fraud similarity with respect to age, gender, qualification and job was presented. And with the proposed similarity measure, high dimensional personal data were calculated to evaluate how similar to the financial fraud. Calculation results show that the actual fraud has rather high similarity measure compared to the average, from minimal 0.0609 to maximal 0.1667.展开更多
Because prior studies find mixed results on the relation between CEOs’pay performance incentives and a firm’s likelihood of financial reporting fraud,we restudy their relationship using innovative research methods.F...Because prior studies find mixed results on the relation between CEOs’pay performance incentives and a firm’s likelihood of financial reporting fraud,we restudy their relationship using innovative research methods.First,we concentrate on incentives from granting options rather than equity-based incentives.Second,we emphasize vested options,disregarding unvested option holdings,and take the logarithm transformation of option incentives.Third,we analyse the impact of option incentives on future financial reporting irregularities.Using this innovative approach as well as a full sample and a matched sample,we find that an increase in executives’option incentives raises the likelihood of financial reporting violations.Moreover,the effect of option incentives on financial reporting fraud is moderated by auditor effort.In addition,we find that another proxy for the measurement of executives’option incentives,namely,the number of vested options by executives,is highly correlated with the CEO’s vested stock option sensitivity.展开更多
Motivated by the Bagging Partial Least Squares(Bagging PLS)and Principal Component Analysis(PCA)algorithms,a novel approach known as Principal Model Analysis(PMA)method is introduced in this paper.In the proposed PMA ...Motivated by the Bagging Partial Least Squares(Bagging PLS)and Principal Component Analysis(PCA)algorithms,a novel approach known as Principal Model Analysis(PMA)method is introduced in this paper.In the proposed PMA algorithm,the PCA and the Bagging PLS are combined.In this method,multiple PLS models are trained on sub-training sets,derived from the training set using the random sampling with replacement approach.The regression coefficients of all the sub-PLS models are fused in a joint regression coefficient matrix.The final projection direction is then estimated by performing the PCA on the joint regression coefficient matrix.Subsequently,the proposed PMA method is compared with other traditional dimension reduction methods,such as PLS,Bagging PLS,Linear discriminant analysis(LDA)and PLS-LDA.Experimental results on six public datasets demonstrate that our proposed method consistently outperforms other approaches in terms of classification performance and exhibits greater stability.Additionally,it is employed in the application of financial statement fraud identification.PMA and other five algorithms are utilized to financial statement fraud which concerned by the academic community,and the results indicate that the classification of PMA surpassed that of the other methods.展开更多
Financial statement fraud refers to malicious manipulations of financial data in listed companies'annual statements.Traditional machine learning approaches focus on individual companies,overlooking the interactive...Financial statement fraud refers to malicious manipulations of financial data in listed companies'annual statements.Traditional machine learning approaches focus on individual companies,overlooking the interactive relationships among companies that are crucial for identifying fraud patterns.Moreover,fraud detection is a typical imbalanced binary classification task with normal samples outnumbering fraud ones.In this paper,we propose a multi-relational graph convolutional network,named FraudGCN,for detecting financial statement fraud.A multi-relational graph is constructed to integrate industrial,supply chain,and accounting-sharing relationships,effectively encapsulating the multidimensional and complex interactions among companies.We then develop a multi-relational graph convolutional network to aggregate information within each relationship and employ an attention mechanism to fuse information across multiple relationships.The attention mechanism enables the model to distinguish the importance of different relationships,thereby aggregating more useful information from key relationships.To alleviate the class imbalance problem,we present a diffusion-based under-sampling strategy that strategically selects key nodes globally for model training.We also employ focal loss to assign greater weights to harder-to-classify minority samples.We build a real-world dataset from the annual financial statement of listed companies in China.The experimental results show that FraudGCN achieves an improvement of 3.15%in Macro-recall,3.36%in Macro-F1,and 3.86%in GMean compared to the second-best method.The dataset and codes are publicly available at:https://github.com/XNetLab/MRG-for-Finance.展开更多
The aim of this paper is to present different aspects of creativity in accounting. Firstly, since creative accounting allows to present a company＇s financial condition in a more favourable light, the ＂advantages＂ o...The aim of this paper is to present different aspects of creativity in accounting. Firstly, since creative accounting allows to present a company＇s financial condition in a more favourable light, the ＂advantages＂ of this creativity are identified. This is followed by the enumeration of negative aspects of creative accounting, to illustrate how creativity in accounting may help to falsify the true picture of a company＇s business. Next, the concept of creative accounting and aggressive accounting （the latter, identified rightly, with a financial and accounting fraud） is explained. Finally, methods that are commonly used to falsify accounts, as well as the premises for and the circumstances in which such criminal practices occur, are presented. The paper draws on the results of other scientific works （books and articles published in scientific journals） in which the main and most interesting views of other authors have been presented. A critical analysis method was employed, followed by a comparison of legal texts and legal acts. The scientific conclusion was conducted using the deduction, and, partially, the induction method. It has been shown that it is justified to differentiate in accounting between creativity that is allowable within the established legal limits and practice based upon the use of creativity in a manner contrary to law. Further, it has been concluded that where creativity in accounting goes beyond the limits unambiguously established by the law, it definitely carries attributes of an offence. It has been also noted that due to continuous changes in the functioning conditions of enterprises, the above mentioned limits cannot be fixed or rigid. Thus it is necessary to control and analyse the creativity of accounting and financial staff working in enterprises in order to adapt the binding regulations to the needs of current markets.展开更多
The detection of anomalous events in huge amounts of data is sought in many domains.For instance,in the context of financial data,the detection of suspicious events is a prerequisite to identify and prevent attempts t...The detection of anomalous events in huge amounts of data is sought in many domains.For instance,in the context of financial data,the detection of suspicious events is a prerequisite to identify and prevent attempts to defraud.Hence,various financial fraud detection approaches have started to exploit Visual Analytics techniques.However,there is no study available giving a systematic outline of the different approaches in this field to understand common strategies but also differences.Thus,we present a survey of existing approaches of visual fraud detection in order to classify different tasks and solutions,to identify and to propose further research opportunities.In this work,fraud detection solutions are explored through five main domains:banks,the stock market,telecommunication companies,insurance companies,and internal frauds.The selected domains explored in this survey were chosen for sharing similar time-oriented and multivariate data characteristics.In this survey,we(1)analyze the current state of the art in this field;(2)define a categorization scheme covering different application domains,visualization methods,interaction techniques,and analytical methods which are used in the context of fraud detection;(3)describe and discuss each approach according to the proposed scheme;and(4)identify challenges and future research topics.展开更多
基金supported by the 2021 Guangdong Province(China)Science and Technology Plan Project“Research and Application of Key Technologies for Multi-level Knowledge Retrieval Based on Big Data Intelligence”(Project No.2021B0101420004)the 2022 commissioned project“Cross-border E-commerce Taxation and Related Research”from the State Taxation Administration Guangdong Provincial Taxation Bureau,China.
文摘Purpose:This study aims to integrate large language models(LLMs)with interpretable machine learning methods to develop a multimodal data-driven framework for predicting corporate financial fraud,addressing the limitations of traditional approaches in long-text semantic parsing,model interpretability,and multisource data fusion,thereby providing regulatory agencies with intelligent auditing tools.Design/methodology/approach:Analyzing 5,304 Chinese listed firms’annual reports(2015-2020)from the CSMAD database,this study leverages the Doubao LLMs to generate chunked summaries and 256-dimensional semantic vectors,developing textual semantic features.It integrates 19 financial indicators,11 governance metrics,and linguistic characteristics(tone,readability)with fraud prediction models optimized through a group of Gradient Boosted Decision Tree(GBDT)algorithms.SHAP value analysis in the final model reveals the risk transmission mechanism by quantifying the marginal impacts of financial,governance,and textual features on fraud likelihood.Findings:The study found that LLMs effectively distill lengthy annual reports into semantic summaries,while GBDT algorithms(AUC>0.850)outperform the traditional Logistic Regression model in fraud detection.Multimodal fusion improved performance by 7.4%,with financial,governance,and textual features providing complementary signals.SHAP analysis revealed financial distress,governance conflicts,and narrative patterns(e.g.,tone anchoring,semantic thresholds)as key fraud indicators,highlighting managerial intent in report language.Research limitations:This study identifies three key limitations:1)lack of interpretability for semantic features,2)absence of granular fraud-type differentiation,and 3)unexplored comparative validation with other deep learning methods.Future research will address these gaps to enhance fraud detection precision and model transparency.Practical implications:The developed semantic-enhanced evaluation model provides a quantitative tool for assessing listed companies’information disclosure quality and enables practical implementation through its derivative real-time monitoring system.This advancement significantly strengthens capital market risk early warning capabilities,offering actionable insights for securities regulation.Originality/value:This study presents three key innovations:1)A novel“chunking-summarizationembedding”framework for efficient semantic compression of lengthy annual reports(30,000 words);2)Demonstration of LLMs’superior performance in financial text analysis,outperforming traditional methods by 19.3%;3)A novel“language-psychology-behavior”triad model for analyzing managerial fraud motives.
基金supported by the Science and Technology Research Project of Chongqing Education Commission(KJQN201801103)the Humanities and Social Science Research Project of Chongqing Education Commission(20SKGH176)the General Funded Projects of Chinese Postdoctoral Science Foundation(2021M693764).
文摘Financial fraud arises from the exaggeration of business interests,and an accurate detection or prediction is a useful tool for both corporate management and capital market systems.A collection of computer technologies has been made on this problem so far,and one of the most important solutions is unsupervised learning algorithms.Among them,most approaches work by analysing the internal relations in financial data and finding a new description of non-fraud firms.However,current studies focus a lot on the geometry attribute of financial data,while overlooking the obvious behaviour patterns and peer effects among firms.This has limited the accuracy of representation and furthermore the detection performance.In this work,a very general class of functions is allowed to represent firms,constraining them by peer effects between firms and presenting an error-distribution-based financial fraud firm detection approach.Experimental results have shown great performance of the proposed approach.
文摘Taking the financial fraud of RX Company in 2020 as a case, this paper collects and studies the investigation report of RX Company, studies the financial fraud means of RX Company from the perspective of investor protection, and analyzes the fraud motivation of RX Company in detail from the perspectives of internal and external causes. Finally, this event triggered thinking on how to implement feasible measures to protect the interests of investors.
文摘As financial criminal methods become increasingly sophisticated, traditional anti-money laundering and fraud detection approaches face significant challenges. This study focuses on the application technologies and challenges of big data analytics in anti-money laundering and financial fraud detection. The research begins by outlining the evolutionary trends of financial crimes and highlighting the new characteristics of the big data era. Subsequently, it systematically analyzes the application of big data analytics technologies in this field, including machine learning, network analysis, and real-time stream processing. Through case studies, the research demonstrates how these technologies enhance the accuracy and efficiency of anomalous transaction detection. However, the study also identifies challenges faced by big data analytics, such as data quality issues, algorithmic bias, and privacy protection concerns. To address these challenges, the research proposes solutions from both technological and managerial perspectives, including the application of privacy-preserving technologies like federated learning. Finally, the study discusses the development prospects of Regulatory Technology (RegTech), emphasizing the importance of synergy between technological innovation and regulatory policies. This research provides guidance for financial institutions and regulatory bodies in optimizing their anti-money laundering and fraud detection strategies.
文摘Luckin Coffee grew rapidly in the past few years and it was the fastest Chinese Concept company go public in NASDAQ.However,its stock price plunged in 2020 due to financial fraud.It is crucial for Luckin Coffee to think about how to recover from the financial fraud and become a business representative of the“new retail”under the Internet model again.To recover from the financial crisis caused and gain profit,Luckin could improve its existing business model by improving product development,strengthening capital management in both marketing and expansion.The internal structure can also be adjusted by diversifying equity and replacing the manager team.Taking Luckin Coffee as an example,this paper analyzes the causes of financial fraud,propose solutions to recover from it,and evaluate the effectiveness of solutions in the context of reality.
文摘Financial fraud,which has become a global issue,is a subject of discussion,surpassing time.Financial fraud significantly undermines investors’confidence and affects the health of capital markets.Hence,it is valuable to explore the reasons for committing financial fraud and propose solutions to this issue.This paper focuses on two financial fraud cases in recent years,Toshiba in 2015 and Luckin Coffee in 2020,analyzes and compares the reasons for the financial fraud in terms of pressure and opportunity factors,as well as proposes comprehensive suggestions for dealing with the corporate financial fraud issue.
基金supported by the National Key R&D Program of China(No.2022YFB4501704)the National Natural Science Foundation of China(Grant No.62102287)the Shanghai Science and Technology Innovation Action Plan Project(Nos.22YS1400600 and 22511100700).
文摘The landscape of financial transactions has grown increasingly complex due to the expansion of global economic integration and advancements in information technology.This complexity poses greater challenges in detecting and managing financial fraud.This review explores the role of Graph Neural Networks(GNNs)in addressing these challenges by proposing a unified framework that categorizes existing GNN methodologies applied to financial fraud detection.Specifically,by examining a series of detailed research questions,this review delves into the suitability of GNNs for financial fraud detection,their deployment in real-world scenarios,and the design considerations that enhance their effectiveness.This review reveals that GNNs are exceptionally adept at capturing complex relational patterns and dynamics within financial networks,significantly outperforming traditional fraud detection methods.Unlike previous surveys that often overlook the specific potentials of GNNs or address them only superficially,our review provides a comprehensive,structured analysis,distinctly focusing on the multifaceted applications and deployments of GNNs in financial fraud detection.This review not only highlights the potential of GNNs to improve fraud detection mechanisms but also identifies current gaps and outlines future research directions to enhance their deployment in financial systems.Through a structured review of over 100 studies,this review paper contributes to the understanding of GNN applications in financial fraud detection,offering insights into their adaptability and potential integration strategies.
基金supported by National Natural Science Foundation of China(61925203)Natural Science Foundation of Xiamen(3502Z20227191)+1 种基金Fundamental Research Funds for the Central Universities(ZQN-1010)Natural Science Foundation of Fujian Province(2023J01137,2021J01316).
文摘Fraud detection is an important task of financial supervision and financial risk management.Several fraud behaviors can be connected with cycle transactions in which the money initially sent from one bank account eventually returns back to the same account.In this paper,we propose an efficient method for detecting such kind of frauds from large-scale financial transaction data.The method first constructs a transaction graph after pre-processing the original data,then divides the graph into its strongly connected components,and finally uses multiple threads to enumerate temporal cycles on different components in a parallel manner.Existing temporal cycle enumeration algorithms usually constraint the length of the cycle or the size of the time-window,which are not suitable for the specific application of financial fraud detection.In light of this,we extend the classical Johnson algorithm to enumerate temporal cycles without length and time-window constraints.To further improve the efficiency of enumeration,we introduce a block-time mechanism that avoids unnecessary multiple explorations of the same parts of the graph components.Experiments show that our method,with multithreading,is on average 15–20 times,and even 100 times faster than the existing competitor.Additionally,we adopt strategies such as amount constraint during cycle enumeration,which assist in reducing the false-positive rate of detected frauds.
文摘Large-scale accounting scandals which were reflected to the world public opinion particularly in the 2000s （such as Enron, Lucent, Xerox, and Parmalat Bank for Reconstruction） carried the matter of fraudulent financial reporting which was made to deceive the financial statement users （Fraudulent Financial Report （FFR）） to the forefront in the agenda of the academicians, operators, and regulatory authorities. As in every crime action, the most effective measure to be taken in preventing FFR events is to try to prevent the FFR before arising. In order to achieve this, in the most effective manner, FFR events should be determined in the formation process. In this study, fraudulent financial statements are tried to be determined by using financial ratios. For this, financial statements of 22 companies which transact in the textile industry in Istanbul Stock Exchange （ISE） were examined. Twenty-three financial ratios were selected for the purpose of determining the risk of fraudulence in the financial statements of the selected companies. These ratios increased in value by multiple regression analysis. The findings which were obtained in the study indicated that some financial statements had the risk of fraudulence. It was concluded that the ratios of inventory/current asset, total debt ratio, and equity turnover rate were a good indicator in the determination of fraudulent financial statements.
文摘By summarizing the factor of the financial statement fraud in existing research outcome, the paper confirms the discriminating characteristic of the financial statement fraud and sets up a theoretic model to discriminate the financial statement fraud. Using radial basis function neural network, regarding to the swatch that the listed company that is punished by the Securities Regulatory Commission or the Ministry of Finance, and according to the clustering elements, validating across by set one aside, the paper verifies respectively the 22 characteristics and 31 characteristics of discriminating model. According to the clustering elements, validating across by set one aside, the paper verifies respectively the 31 characteristics and 8 characteristics selected by Fisher-ratio of discriminating model. The research outcome indicates the discriminating ability of the model including 8 characteristics is better elevated than the traditional model including 31 characteristics by comparing the disciplinary error and the forecast precision.
文摘This paper studied the ＂pressure＂ and ＂opportunity＂ factors that caused financial statement fraud on the basis of the data of 41 A-share listed companies （1999-2004） in China which had been forfeited by China Securities Regulatory Commission （SRC） because of accounting irregularities. The author found that avoiding ＂ST＂ and ＂PT＂ was the primary pressure, and the opportunities mainly came from the higher top 10 shareholders＇ ownership concentration, the lower proportion of independent directors, the fewer number of directorate meetings and shares owned by the directorate members, board chairman and CEO held by one person and the ineffective supervisor boards. We also found that the companies involved financial statement fraud had the lower first majority shareholder＇s share proportion and they changed CPA firm more frequently.
文摘To evaluate the applicability of the M-score model in the Chinese capital market,this research observed 190 financial fraud samples punished by the China Securities Regulatory Commission(CSRC)in the years from 2014 to 2018.The test results indicate that two types of errors are high,which means that the applicability of the M-score is unacceptable.Therefore,in this paper,a 9-index model is constructed by Wald's backward stepwise regression method,and the optimal threshold is set by the Beneish expected cost method(ECM).The accuracy of the modified M-score is significantly improved,especially the Type I error rate of is reduced from 70.37%to 19.75%.The receiver operating characteristic(ROC)curve test also proves the superior identification effect of the modified M-score applied in the Chinese market.Finally,variables such as current ratio,fixed asset index,and equity concentration in the modified model could represent the fraud characteristics of Chinese listed companies.
文摘Understanding the root causes of fraud and learning about the most effective fraud prevention mechanisms are critical in reducing the incidence of financial fraud. Therefore, this study solicits the views of fraud investigators on the existence and effectiveness of fraud prevention mechanisms within Government-Linked Companies （GLCs）. Specifically, effective fraud prevention mechanisms （FPM） should be undertaken by five critical groups, namely the board of directors, audit committees, external auditors, internal auditors, and anti-fraud specialists. A total of 147 questionnaires were distributed to fraud investigators of all public listed GLCs in Malaysia. Out of those, a total of 45 usable responses were received, which represents an effective response rate of 30.6%. In terms of existence, the respondents ranked ＂management review of internal controls＂ and ＂external audits of financial statements＂ as the top-most FPMs. This was followed by other mechanisms such as operational audits, internal audits and internal control review/improvements by departments. Out of the 27 fraud prevention mechanisms, fraud investigators perceiyed surprise audits, fraud hotline, fraud prevention program and training, anti-fraud policy, fraud vulnerability reviews, operational audits, whistle-blowing policy, internal audit or fraud examination department, and, imposing penalty and disciplinary action as more effective than the others. Overall, this study provides important insights to practitioners and organizations in identifying fraud prevention mechanisms that are most effective.
文摘Different from foreign capital markets,china’s domestic capital markets are special,which also determines that the research on financial pressure starts from the reform of state-owned enterprises,and draws lessons from the relevant theories of financial risk and financial fraud,thus gradually forming a more diversified research results.
基金Project(RDF 11-02-03)supported by the Research Development Fund of XJTLU,China
文摘Information analysis of high dimensional data was carried out through similarity measure application. High dimensional data were considered as the a typical structure. Additionally, overlapped and non-overlapped data were introduced, and similarity measure analysis was also illustrated and compared with conventional similarity measure. As a result, overlapped data comparison was possible to present similarity with conventional similarity measure. Non-overlapped data similarity analysis provided the clue to solve the similarity of high dimensional data. Considering high dimensional data analysis was designed with consideration of neighborhoods information. Conservative and strict solutions were proposed. Proposed similarity measure was applied to express financial fraud among multi dimensional datasets. In illustrative example, financial fraud similarity with respect to age, gender, qualification and job was presented. And with the proposed similarity measure, high dimensional personal data were calculated to evaluate how similar to the financial fraud. Calculation results show that the actual fraud has rather high similarity measure compared to the average, from minimal 0.0609 to maximal 0.1667.
基金financial support from the National Natural Science Foundation of China(Grant No.71620107005)the 111 Project“Innovation and Talents Base of Financial Security and Development”(Grant No.B18043)support from the Chinese National Science Foundation(No.71672149 and No.71972157)
文摘Because prior studies find mixed results on the relation between CEOs’pay performance incentives and a firm’s likelihood of financial reporting fraud,we restudy their relationship using innovative research methods.First,we concentrate on incentives from granting options rather than equity-based incentives.Second,we emphasize vested options,disregarding unvested option holdings,and take the logarithm transformation of option incentives.Third,we analyse the impact of option incentives on future financial reporting irregularities.Using this innovative approach as well as a full sample and a matched sample,we find that an increase in executives’option incentives raises the likelihood of financial reporting violations.Moreover,the effect of option incentives on financial reporting fraud is moderated by auditor effort.In addition,we find that another proxy for the measurement of executives’option incentives,namely,the number of vested options by executives,is highly correlated with the CEO’s vested stock option sensitivity.
基金Supported by the Beijing Municipal Social Science Foundation(SZ202210005004)Beijing Natural Science Foundation(9242004)。
文摘Motivated by the Bagging Partial Least Squares(Bagging PLS)and Principal Component Analysis(PCA)algorithms,a novel approach known as Principal Model Analysis(PMA)method is introduced in this paper.In the proposed PMA algorithm,the PCA and the Bagging PLS are combined.In this method,multiple PLS models are trained on sub-training sets,derived from the training set using the random sampling with replacement approach.The regression coefficients of all the sub-PLS models are fused in a joint regression coefficient matrix.The final projection direction is then estimated by performing the PCA on the joint regression coefficient matrix.Subsequently,the proposed PMA method is compared with other traditional dimension reduction methods,such as PLS,Bagging PLS,Linear discriminant analysis(LDA)and PLS-LDA.Experimental results on six public datasets demonstrate that our proposed method consistently outperforms other approaches in terms of classification performance and exhibits greater stability.Additionally,it is employed in the application of financial statement fraud identification.PMA and other five algorithms are utilized to financial statement fraud which concerned by the academic community,and the results indicate that the classification of PMA surpassed that of the other methods.
基金supported by the National Natural Science Foundation of China(Nos.62272379,T2341003,U22B2019,and 62102310)the Natural Science Basic Research Plan in Shaanxi Province(No.2021JM-018)+1 种基金the Key R&D in Shaanxi Province(No.2023-YBGY-269)the Fundamental Research Funds for the Central Universities(No.xzy012023068).
文摘Financial statement fraud refers to malicious manipulations of financial data in listed companies'annual statements.Traditional machine learning approaches focus on individual companies,overlooking the interactive relationships among companies that are crucial for identifying fraud patterns.Moreover,fraud detection is a typical imbalanced binary classification task with normal samples outnumbering fraud ones.In this paper,we propose a multi-relational graph convolutional network,named FraudGCN,for detecting financial statement fraud.A multi-relational graph is constructed to integrate industrial,supply chain,and accounting-sharing relationships,effectively encapsulating the multidimensional and complex interactions among companies.We then develop a multi-relational graph convolutional network to aggregate information within each relationship and employ an attention mechanism to fuse information across multiple relationships.The attention mechanism enables the model to distinguish the importance of different relationships,thereby aggregating more useful information from key relationships.To alleviate the class imbalance problem,we present a diffusion-based under-sampling strategy that strategically selects key nodes globally for model training.We also employ focal loss to assign greater weights to harder-to-classify minority samples.We build a real-world dataset from the annual financial statement of listed companies in China.The experimental results show that FraudGCN achieves an improvement of 3.15%in Macro-recall,3.36%in Macro-F1,and 3.86%in GMean compared to the second-best method.The dataset and codes are publicly available at:https://github.com/XNetLab/MRG-for-Finance.
文摘The aim of this paper is to present different aspects of creativity in accounting. Firstly, since creative accounting allows to present a company＇s financial condition in a more favourable light, the ＂advantages＂ of this creativity are identified. This is followed by the enumeration of negative aspects of creative accounting, to illustrate how creativity in accounting may help to falsify the true picture of a company＇s business. Next, the concept of creative accounting and aggressive accounting （the latter, identified rightly, with a financial and accounting fraud） is explained. Finally, methods that are commonly used to falsify accounts, as well as the premises for and the circumstances in which such criminal practices occur, are presented. The paper draws on the results of other scientific works （books and articles published in scientific journals） in which the main and most interesting views of other authors have been presented. A critical analysis method was employed, followed by a comparison of legal texts and legal acts. The scientific conclusion was conducted using the deduction, and, partially, the induction method. It has been shown that it is justified to differentiate in accounting between creativity that is allowable within the established legal limits and practice based upon the use of creativity in a manner contrary to law. Further, it has been concluded that where creativity in accounting goes beyond the limits unambiguously established by the law, it definitely carries attributes of an offence. It has been also noted that due to continuous changes in the functioning conditions of enterprises, the above mentioned limits cannot be fixed or rigid. Thus it is necessary to control and analyse the creativity of accounting and financial staff working in enterprises in order to adapt the binding regulations to the needs of current markets.
基金The research leading to these results has received funding from the Centre for Visual Analytics Science and Technology(CVAST),funded by the Austrian Federal Ministry of Science,Research,and Economy in the exceptional Laura Bassi Centres of Excellence initiative(#822746).
文摘The detection of anomalous events in huge amounts of data is sought in many domains.For instance,in the context of financial data,the detection of suspicious events is a prerequisite to identify and prevent attempts to defraud.Hence,various financial fraud detection approaches have started to exploit Visual Analytics techniques.However,there is no study available giving a systematic outline of the different approaches in this field to understand common strategies but also differences.Thus,we present a survey of existing approaches of visual fraud detection in order to classify different tasks and solutions,to identify and to propose further research opportunities.In this work,fraud detection solutions are explored through five main domains:banks,the stock market,telecommunication companies,insurance companies,and internal frauds.The selected domains explored in this survey were chosen for sharing similar time-oriented and multivariate data characteristics.In this survey,we(1)analyze the current state of the art in this field;(2)define a categorization scheme covering different application domains,visualization methods,interaction techniques,and analytical methods which are used in the context of fraud detection;(3)describe and discuss each approach according to the proposed scheme;and(4)identify challenges and future research topics.
