資訊管理學報
陳育仁;
頁: 75-101
日期: 2022/01
摘要: 對於投資大眾來說，最關注其投資的企業是否發生舞弊之現象；而企業最常見的舞弊手法包括財報不實、資產掏空與內線交易等，其中又以資產掏空所造成的後果最為嚴重，可能導致整個企業停擺，使得投資者血本無歸。因此，如何有效的偵測企業掏空與否，已成為舞弊審計重要的課題之一。 本研究主要考量企業財務結構、償債能力、經營能力、獲利能力、現金流量與成長力等財務性指標以及公司治理方面之股權結構、董監事組成、關係人交易與管理型態等非財務性指標，並經由主成分分析與逐步迴歸進行資產掏空偵測指標之建立，再整合支援向量機與改良式基因演算法進行企業資產掏空之偵測，以提供投資大眾選擇目標企業投資時之決策參考，進而降低投資者之投資風險。針對上述目的，本研究主要研究項目包括: (i)資產掏空偵測特徵指標之建立，(ii)資產掏空偵測方法之發展以及(iii)資產掏空方法之驗證與評估。
關鍵字: 資產掏空;舞弊偵測;主成分分析;逐步迴歸;支援向量機;改良式基因演算法;
Abstract: Management fraud is a major concern among investors. Misstated financial statements, asset misappropriation, and insider trading are common forms of fraud by enterprises. Among these, asset misappropriation presents the highest risk for substantial losses as it may result in corporate shutdown and the loss of the life savings of investors. Therefore, identifying methods to effectively detect bad management at the earliest and prevent corporate asset misappropriation is critical in the study of fraud audits. This study considers several financial structure indicators - solvency, operating capacity, profitability, cash flow, and growth ability. It also examines several non-financial indicators - shareholding structure, board composition, related party transactions, and management style in corporate governance. The feature indicators for asset misappropriation detection are first established through principal component analysis and stepwise regression. Support vector machine (SVM) and queen genetic algorithm (QGA) are then combined to effectively detect corporate asset misappropriation, providing a reference to investors and creditors for investment decision-making and thereby reducing their investment risks. This objective is achieved by (i) establishing feature indicators for asset misappropriation detection, (ii) developing an asset misappropriation detection method, and (iii) demonstrating and evaluating the proposed method.
Keywords: Asset misappropriation, fraud detection, principal component analysis, stepwise regression, support vector machine, queen genetic algorithm ;
瀏覽次數: 11517 下載次數: 577
引用 導入Endnote
頁: 75-101
日期: 2022/01
摘要: 對於投資大眾來說，最關注其投資的企業是否發生舞弊之現象；而企業最常見的舞弊手法包括財報不實、資產掏空與內線交易等，其中又以資產掏空所造成的後果最為嚴重，可能導致整個企業停擺，使得投資者血本無歸。因此，如何有效的偵測企業掏空與否，已成為舞弊審計重要的課題之一。 本研究主要考量企業財務結構、償債能力、經營能力、獲利能力、現金流量與成長力等財務性指標以及公司治理方面之股權結構、董監事組成、關係人交易與管理型態等非財務性指標，並經由主成分分析與逐步迴歸進行資產掏空偵測指標之建立，再整合支援向量機與改良式基因演算法進行企業資產掏空之偵測，以提供投資大眾選擇目標企業投資時之決策參考，進而降低投資者之投資風險。針對上述目的，本研究主要研究項目包括: (i)資產掏空偵測特徵指標之建立，(ii)資產掏空偵測方法之發展以及(iii)資產掏空方法之驗證與評估。
關鍵字: 資產掏空;舞弊偵測;主成分分析;逐步迴歸;支援向量機;改良式基因演算法;
Fraud Detection for Corporate Asset Misappropriation
Abstract: Management fraud is a major concern among investors. Misstated financial statements, asset misappropriation, and insider trading are common forms of fraud by enterprises. Among these, asset misappropriation presents the highest risk for substantial losses as it may result in corporate shutdown and the loss of the life savings of investors. Therefore, identifying methods to effectively detect bad management at the earliest and prevent corporate asset misappropriation is critical in the study of fraud audits. This study considers several financial structure indicators - solvency, operating capacity, profitability, cash flow, and growth ability. It also examines several non-financial indicators - shareholding structure, board composition, related party transactions, and management style in corporate governance. The feature indicators for asset misappropriation detection are first established through principal component analysis and stepwise regression. Support vector machine (SVM) and queen genetic algorithm (QGA) are then combined to effectively detect corporate asset misappropriation, providing a reference to investors and creditors for investment decision-making and thereby reducing their investment risks. This objective is achieved by (i) establishing feature indicators for asset misappropriation detection, (ii) developing an asset misappropriation detection method, and (iii) demonstrating and evaluating the proposed method.
Keywords: Asset misappropriation, fraud detection, principal component analysis, stepwise regression, support vector machine, queen genetic algorithm ;
瀏覽次數: 11517 下載次數: 577
引用 導入Endnote