The Research Of Private Enterprises Credit Scoring Based On Multivariate Adaptive Regression Splines

Private enterprises are the important part of our country economics. They play an important role in the aspects such as providing job opportunities, promoting orientation of schemes and technique, and promoting economics development. However, the live environment and development environment of private enterprises is not good. They meet big difficulties in many aspects, especially in the financing aspect. There are many reasons contributing to the financing difficulties. From the enterprise aspect, some private enterprises have bad credit conditions, high risks, less capital, and opaque credit environment. From the bank aspect, some business banks have not enough marketing ideas, and they can not apply to private enterprises'need in the aspects such as organizations, credit scoring, debt authority, and inner management. Therefore, it has the important theoretical and practical significance to establish credit scoring system and credit methods according to private enterprises characters, draft scientific realistic private enterprises'credit scoring schemes, and rate private enterprises credit rank objectively.This article is from the national social science fund project"Research on private enterprises credit scoring index systems and credit scoring methods". It is divided into six chapters.The first chapter is the introduction. It introduces the article's background, brief work, orientation, and the whole structure. The orientation of the article is that we build the credit scoring system according to the private enterprises'finance character and it is the first time that we apply Multivariate Adaptive Regression Splines to enterprises'credit scoring.The second chapter is summarizing of the enterprises'credit scoring. It indicates the concept of credit scoring, the meaning of credit scoring, and the Economics theory of credit scoring (including Information Asymmetry Theory, Expenditure Business theory, and Game Theory). And it expatiates the orientation of important meaning of credit scoring in Economics aspect.The third chapter is the literature review of credit scoring methods. It introduces three credit scoring methods: Discriminant Analysis, Logistic Regression and Artificial neural networks. Then it analyzes and compares three credit scoring methods'advantages and disadvantages.The fourth chapter is the theory of the Multivariate Adaptive Regression Splines model-the theoretical core of the whole article. This chapter first introduces the development of this model and its advantages. Then it discusses the model building, the theory basis of variables'choice, model decomposing, and model forecasting thoroughly, which build a strong theory base for the empirical study.The fifth chapter is the empirical study-the core of the whole article. It forecasts whether the private enterprise will default by establishing the Multivariate Adaptive Regression Splines model. First, it chooses 194 private enterprise samples by excluding the outliners. Second, using MARS software calculating the contribution rate, it chooses the net assets per share, cash ratio, net assets rate of return, net profit ratio of main business, and debt to assets ratio as the indexes. Third, it applies the Multivariate Adaptive Regression Splines model to forecast. The correct rate of the forecasting is 95.9%. The correct rate of discriminating the second type error reaches to 80%. Both of the correct rates are higher than the Discriminant Analysis and Logistic Regression, which could prove that MARS has a good practice value to discriminate private enterprises'default condition.The sixth chapter is the conclusion. The limit of the article is the lack of private enterprise samples, not excluding the influence of industry factors and not determining the nature of scoring. In future, the article should enlarge the frame of the range of the samples, keep further study of the index choosing, and improve the correct rate of the classification.It is the first time that we have applied Multivariate Adaptive Regression Splines model to private enterprises'credit scoring in our country. Therefore, it has a positive meaning in establishing the theory of private enterprises'credit scoring system and credit scoring methods. Also, it prompts the improving of the private enterprises'financing condition, prompts the reform of bank loan system, and to some extant prompts the whole national economics development.