Combined Algorithm for Detection of Manipulation in Financial Statements

Abstract

This paper identifies several approaches to detection of manipulation in financial statements and creates a combined practical approach, which uses each of the existing algorithms in the complement manner to the others. It also introduces new P-Score formula, which in conjunction with Altman Z-score formula creates new approach to detection of fraud in financial statements of the public companies. It uses data from multiple yearly statements of the company in the sample to monitor manipulation signs. The analysis, conducted on the sample of companies, which were charged with fraud, shows that new P-Score/Z-Score approach gives 82.76% chance of detecting fraud and combined approach gives 96.55% chance of detecting fraud in financial statements, which is a significant improvement over other models. The paper also introduces a finite set of variables, which can be found in the financial statements of the public companies in easy and transparent manner. The ultimate goal of the study is to create a basis for using XBRL based data to detect financial statement fraud automatically.