Credit Risk Rating Of Small Industry Enterprise Based On Default Risk Discrimination

The essence of credit rating is revealing the relationship between loan data and default risk and ascertaining default probability of loan. It is pressing to build credit rating system applied to small industrial enterprises for financial institutions because there are small number of credit rating literature referring small industrial enterprises which have small firm size and incomplete financial information.Study of small industrial enterprises credit rating system based on discriminatory power consists of three parts. The first part builds credit rating indicator system which can discriminate small industrial enterprises default state significantly and reflect small industrial enterprises customer's loan repayment ability on the basis of indicator's influence on model discriminate accuracy. The second part builds credit rating model in accordance with the principle the higher scores of non-default enterprises, the lower scores of default enterprises. This paper utilizes multi-objective programming model to weight selected indicators and builds credit scoring model. The last part is credit rating division. This research constructs nonlinear programming model to divide the credit rating according to the LGD pyramid and maximum credit difference degree. The division can meet the pyramid standard that customers with lower LGD should be divided into higher level and ensure customers with different credit status are more likely to be divided into different credit level. This paper includes five chapters. In chapter 1, we analyze the topic basis, the relative research development, methods and so on. In chapter 2, we discuss the theory of credit rating based on discriminatory power. In chapter 3, we construct indicator system of credit rating for small industrial business based on Fisher discrimination. In chapter 4, we establish credit scoring model based on the maximizing default identification ability of combination weighting. In chapter 5, we study credit rating division model based on the largest credit difference degree. Chapter 6 is the conclusions and outlook. The main works and innovations of this paper are as follows:(1)The main works and innovations of credit rating division:Firstly, we build up a nonlinear programming model to divide the credit rating with the objective function that the sum of 8 first-order differential of credit score is maximum, which ensure customers with different credit status are more likely to be divided into different credit level. Secondly, we construct a nonlinear programming model to divide the credit rating with the constraint that the LGD is strictly increasing with credit rating from high to low, which can meet the pyramid standard that customers with lower LGD should be divided in higher level, we can avoid the unreasonable phenomenon that customers with higher LGD while the credit level not low.(2)The main works and innovations of combination weighting:By minimizing the distance of non-default enterprises' weighted data to positive ideal point as the first objective function, minimizing the distance of default enterprises' weighted data to negative ideal point as the second objective function, multi-objective programming model is constructed to solve the optimal combination weights. The optimal combination weights satisfy the goal "the higher scores of non-default enterprises, the lower scores of default enterprises" and ensure that credit rating model maximizes the distance of non-default enterprises and default enterprises. It changes the disadvantage that combination weights is contrary to credit rating purpose in existing research, and changes the disadvantage that a lot of the score overlap between non-default and default enterprises and the ability of distinguish between non-default and default enterprises is low in existing research.(3)The main works and innovations of indicator selection:According that in two kinds of condition with or without specific indicators, effect of using Fisher discrimination analysis on identification accuracy of default state reflects influence of specific indicator to the default state, we delete indicators without effect on identification accuracy of default state, and remain indicators with significant effect on identification accuracy of default state, which perfect the disadvantage of most existing research that the standard of selecting indicator has no relationship with default state.