Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Faculty of Business

First Advisor

Professor Gary S. Munroe


A contemporary issue of concern to both external auditors and financial statement users is fraud-detection by auditors. The ability of auditors to detect material irregularities, including fraud, should be enhanced to enable them to apply "reasonable skill and care" in carrying out the audit. Such proficiency in fraud detection is needed if the profession is to avoid costly litigation, ever-increasing indemnity insurance and erosion of the profession's credibility. The thesis maintains that such enhancement can be achieved if auditors both utilise knowledge about the aetiology of fraud in psychology, sociology and criminology as well as by synthesising a broad range of approaches to fraud detection. The multidisciplinary discussion of the aetiology of fraud enabled the development of a three-component model. The model's three components are: rationalisations (R), opportunity (0) and a crime-prone motivated person (P), hence the acronym ROP. Next, a close examination of relevant auditing guidelines and a number of fraud detection models that have been proposed were used to develop an eclectic fraud detection model (with the ROP model as one of its components). The applicability of the ROP model was determined in a study of 50 major fraud cases investigated and prosecuted by the Major Fraud Group (MFG) of the Victoria police. The study identified a number of inter-relationships between offence, offender and victim characteristics. The findings obtained also confirmed the applicability of the model in the field and yielded a two-level criminal profile of serious fraud offenders which includes a new taxonomy of such offenders. The taxonomy consists of twelve specific typologies. In addition, the MFG study findings cast doubt (I) on Gottfredson and Hirschi's (1990) assertion in their General Theory of Crime that white-collar offenders are not significantly different from common offenders and (2) on a basic premise of Loebbecke et al.'s (1989) fraud risk-assessment model that all three components of their model need to be present for fraud to occur. The experience of auditors with detecting six different types of material irregularities, including management fraud, employ fraud and error, was investigated in a postal survey of 108 auditors. The findings provide support for the applicability of the eclectic fraud detection model. The survey also found that: it is rare for even experienced auditors to encounter material irregularities; that different types of irregularity (e.g., management fraud) occur more frequently in some industries (manufacturing and construction) than in others; the irregularity is likely to take one form (e.g., window dressing and misappropriation of funds) rather than another; and management review and tests of controls are more likely to alert an auditor to the existence of management fraud. In support of earlier research findings, data analysis revealed that the lack of an effective internal control system and the absence of a code of corporate conduct are statistically significant correlates of an irregularity having a material impact on the financial accounts of a company. In contrast to claims by Loebbecke et al. (1989), the survey findings show that fraud risk-assessment utilising red flags alone is not effective and the presence of only two (and not all three) of their model's components need to be present for management fraud to occur. Both the ROP model and the eclectic fraud detection model were further refined in the light of the findings from the two empirical studies. Without ignoring limitations of the two surveys, the work reported in the present thesis sheds new light on the aetiology of fraud, shows that neither audit experience nor red flags alone is sufficient to improve auditors' fraud detection performance and provide another dimension to fraud risk- assessment. The new knowledge should be added to the auditor's armoury to enhance the audit effectiveness and efficiency and to reduce the fraud detection component of the expectation gap.