Assessing Managers Fraud Through Analysis of Board of Directors Report by Data Mining

Abstract

Detecting, evaluating and understanding fraud reports, called as misstated reports, has a long history in the accounting and financial literature. Shareholders choose managers as their agents in the company, so it is usual for them to be sensitive to the honesty of managers’ reports. Also auditors in their investigations are required to evaluate and measure the fraud risk in the methods implemented by the management of entity, and then, to estimate the parameters of audit tests, and their proposed auditing fees. This study has used non-financial approach to detecting and evaluating managers' fraud risk that is based on the analysis of the text of board of directors' report. In this method, the words of the board's reports were reviewed, and after some refinements, a model was presented for evaluating high fraud risk index in companies, using a certain type of regression model, called LASSO. This model is able to identify high fraud risk index in companies, correctly with precision of 89% to 91%.

Keywords

• fraud risk
• fraudulent reporting
• data mining
• lasso regressions