Application of Neural Networks in Identifying Fraudulent Companies with Special Reference to the Software Industry in India

Year 2009 will remain etched in the history of Financial Statement Fraud (FSF) since the Satyam fraud of $1.47 billion in software industry took place that year in India. FSF ultimately leaves the investors deprived. Hence, today's investors need a tool or technique that can help them to find out whether a particular company is fraudulent or not. The technique discussed in this paper is called Neural Network (NN). This study particularly demonstrates the power of the technique which can accurately classify the companies into fraudulent and non-fraudulent on the basis of publicly available information. Two basic objectives are achieved in this study. A total of 33 variables in the form of financial ratios are found after reviewing past literature. To identify the significant factors, data reduction technique, i.e., factor analysis, is applied in SPSS 21. After applying factor analysis, seven factors are found enveloping all the other variables 100%. Assuming the seven factors as input vectors, the Multilayer Perceptron (MLP) method of NN technique is applied in SPSS 21. The screenshots are also displayed in the study to explain the technique more accurately