What blows the whistle on fraud? — Analysis on the risk indicators to detect corporate fraud in post-financial crisis era

Abstract

This study aims to cast a light on the importance of fraud risk management control after the 2007-2008 financial crisis. Apart from accounting statement frauds, the other common types of corporate frauds including asset misappropriation and corruption are also of our interest. The development of the hypotheses in this paper is based on Fraud Diamond Theory, supplemented with Earnings Management and Corporate Social Responsibility perspectives. By examining a plenty of risk indicators from these approaches suggested in the previous fraud-detecting studies, this study intends to refine the conventional fraud-detecting models in order to better predict the occurrence of fraud at earlier stages.

This study firstly conducts descriptive statistic, including mean/median comparison tests such as independent T test or Mann-Whitney U test, for the samples in order to give a fundamental picture of the nature of the dataset. Our total samples include 104 fraud-involved firms out of 3298 firms, which can be further expressed as 1011 firm-quarterly financial information from fraud-involved firms out of 13149 firm-quarterly financial information, and 45 pre-fraud-firm-quarterly financial information out of 13149 firm-quarterly financial information. Then two binary logistic regression models are built to respectively identify: (1) the existence of fraudulent firms based on all quarterly information, and (2) the existence of fraudulent firms based on quarterly information prior to the fraud occurrence. Overall, this study aims to: (1) detect the occurrence of fraud by using the risk indicators from Fraud Diamond Theory, Earnings Management and Corporate Social Responsibility; and (2) further construct updated regression models by using stepwise backward LR method in order to achieve higher reliability and better power of explanatory. In the end, we purpose our outcomes to help external stakeholders to both detect the occurrence of fraud and to recognise fraud in the quarters prior to the announcement dates of fraud enforcement actions in order to achieve higher time value.

The results of this study first indicate that Fraud Diamond Theory, Earnings Management and Corporate Social Responsibility approach together can partly explain the occurrence of fraud and signify the existence of fraud in the pre-fraud quarters. Although suffering from several limitations and still being improvable, this research provides substantial evidences to help future researchers to establish fraud-detecting models and its results also point out quarterly information can better predict the occurrence of fraud than yearly information.