The litigation problem: Analytical procedures for detecting fraudulent financial statement

This dissertation contributes analytical procedures for detecting fraudulent financial statements. The auditing profession is facing a litigation crisis and needs additional research into ways of limiting audit risk due to fraudulent financial statements. Management fraud is responsible for many of the costly litigation cases involving the auditing profession. Therefore, techniques able to detect management fraud will decrease future litigation. To find these effective analytical procedures, this dissertation integrates standard classification techniques with Artificial Neural Networks (ANN's).; The dissertation uses two samples, (1) consisting of proprietary data used in prior works (Bell et al. 1993) and (2) a matched sample. The matched sample includes cases selected from the Securities and Exchange commission's enforcement releases. Using several of the discrimination techniques available two discriminant functions are developed for use in detecting management fraud.; In addition, the dissertation contributes to the literature by assessing the effectiveness of various classification methods. The contributions of certain Artificial Neural Networks as analytical review aids are measured in relation to more standard methods such as logistic regression and discriminant analysis. Specifically, evolutionary ANN's were validated in the dissertation.