Prediction of bond ratings using statistical and neural networks techniques

Abstract

In this research work, an attempt has been made to develop a comprehensive statistical and neural networks models lo predict the ratings of Indian manufacturing sector firms based on their financial characteristics. This report is structured as follows: Chapter 2 provides an outline on the fundamentals of Debt Instruments and Credit Rating. It discusses the concepts of Debt Instruments, Credit Rating, a brief about the Rating Symbols, the Credit Rating Process and the Key Factors of Rating. Chapter 3 mainly outlines the Literature Review on Bond Rating and the previous research that has been carried out in related areas. Chapter 4 broadly discusses the techniques that were adopted for imbibing data, its source, and the factors governing the selection of variables. Chapter 5 consists of the details related to the preprocessing of data, which mainly include removal of outliers, Normalization and Standardization of data, Correlation Analysis and Principal Component Analysis. Chapter 6 describes the Multiple Discriminant Analysis along with a discussion on the Assumptions, Method and Algorithm that were used for the purpose of designing the model. Chapter 7 furnishes an overview about Multinomial Logistic Regression Model and the various steps involved in the process of model building and the strategies that were adopted. Chapter 8 discusses the usage of Neural Network Techniques for Bond Rating Classification. Chapter 9 concludes the research work with the Summary Results, Model Comparison and Conclusion