A Generalized Qualitative-Response Model and the Analysis of Management Fraud

Management fraud has become a topic of increasing interest to the public accounting profession. Prior research indicates that management fraud is seldom experienced by audiGtors. As a result, it is doubtful that auditors have a well-developed cognitive model for making fraud risk assessments as part of the audit planning process. Early research studies attempted to identify factors that could be linked to the occurrence of management fraud, while more recent work has attempted to build models to predict the presence of management fraud. In this paper, we report on a study that uses a powerful generalized qualitative-response model, EGB2, to model and predict management fraud based on a set of data developed by an international public accounting firm. The EGB2 specification includes the probit and logit models and others as special cases. Moreover, EGB2 easily accommodates asymmetric costs of type I and type II errors. This is important for public accounting firms since failure to predict fraud when it is present (a type II error) is usually very costly to the firm in terms of litigation. The results demonstrate good predictive capability for both symmetric and asymmetric cost assumptions.

• 저널Management Science · 42(7) · 1022–1032
• 토픽Imbalanced Data Classification Techniques · Artificial Intelligence
• DOI10.1287/mnsc.42.7.1022
• 저자James V. Hansen, James B. McDonald, William F. Messier, Timothy B. Bell