This study aims to classify the numbers used in recent financial statement, corruption and asset misappropriation fraud schemes in such a way that these classes can be used to design effective proactive analytics-based fraud detection tests.
The data sources for the classification scheme include the court records of fraud prosecutions, investigative reports and research papers related to fraud cases.
Fraudulent numbers are most often amounts that are round, have a strong period-over-period growth, are just above or below internal control thresholds or other targets, are deviations from Benford’s Law, are purposeful duplicates of authentic transactions, are outliers due to being excessively large and are excessively rounded up or down. The study includes several examples of fraudulent numbers.
The fraudulent number types are based on a sample of fraud-related court documents, and the sample might not be representative of the population of detected and undetected frauds. Further research is needed into the detection of corruption/bribery schemes.
The results are important for auditors and forensic accountants running proactive fraud detection tests. The discussions emphasize that the analysis should include refining and rerunning the tests, and then using groupings and filtering to deal with false positives. The importance of an effective audit of the notable transactions is stressed in the concluding section.
The study is an original in-depth coverage of the patterns found in fraudulent numbers. The discussion sections review implementation issues and considerations for future research.
