What are the common methods of enterprise credit rating?

Discriminant analysis is to divide the known defaulting and non-defaulting enterprises into three populations, find out a discriminant function from the characteristics of these populations, and judge which population any observation vector should belong to, test whether there are significant differences between two or more matrices in the measured index variables, and if so, point out which indicators.

1968, Altman took the lead in applying discriminant analysis to financial analysis, enterprise bankruptcy and credit risk analysis, and established the following famous linear discriminant analysis model:

z = 0.0 12x 1+0.0 14x 2+0.033 x3+0.006 x4+0.999 X5，

Where X 1 is current capital/total assets, X2 is retained income/total assets, X3 is interest and tax income/total assets, X4 is equity market value/book value of total liabilities, and X5 is sales income/total assets. The critical value is 2.675. If z is less than the critical value, the borrower is classified as a default group and has a low credit rating. On the contrary, it is classified as a normal group with a high credit rating. When the score is between 1.8 1 and 2.99, Altman finds that the judgment error is large and the repeated area is gray.

Although the linear discriminant analysis model represented by Z model is very suitable for credit rating, there are some problems in this method: (1) the restrictions are too strict, such as requiring the sample data to obey multivariate normal distribution and the covariance matrix to be the same; (2) The model mainly considers financial factors, without considering the influence of non-financial factors such as industry characteristics, enterprise scale and management level; (2) The model is based on historical data, and the prediction of future development is not enough.

Comprehensive evaluation method

Comprehensive evaluation method is a general evaluation of things or phenomena affected by various factors, that is, according to given conditions, each object is given a real number, and a comprehensive score is obtained by other calculation methods such as total score method or weighted average, and then its priority is evaluated accordingly.

Judging from the nature of credit rating itself, enterprise credit rating is an uncertain and vague problem. Therefore, the development trend of comprehensive evaluation method is enough to combine fuzzy theory with enterprise credit rating, thus making the rating results more scientific and accurate.

Artificial neural network method

The so-called artificial neural network is an information processing system or computer based on imitating the structure and function of biological brain, which is called artificial neural network for short. The basic framework of artificial neural network is to imitate biological nerve cells, which are divided into input layer, hidden layer and output layer. Each layer of color includes several points representing processing units. The nodes in the input layer are responsible for receiving external information different from that input by the human brain (as shown in figure 1). The input information received by artificial neural network is the quantized information of various variables, and one input variable corresponds to one input node. The nodes in the hidden layer are responsible for processing the information transmitted from the input layer and converting it into intermediate results to be transmitted to the output layer. The node of the output layer compares the information of the hidden layer with the threshold value to get the final result of the system and output it.

Compared with traditional statistical methods, artificial neural network has the following characteristics: (1) has the ability of self-organization and learning; (2) It can describe the nonlinear relationship between variables in the input data; (3) It can be dynamically adjusted according to the changes of samples and environment, because there is often a nonlinear relationship between various financial indicators of enterprises and credit risk. Therefore, artificial neural network is more suitable for enterprise credit evaluation.

Fuzzy analysis method

Traditional mathematical or statistical methods are based on precise assumptions, but there are many ambiguities or uncertainties in natural science, social science and engineering technology. Human cognitive mode, thinking mode and even reasoning logic also involve many uncertainties. Therefore, traditional methods can't solve this kind of uncertainty problem, while fuzzy mathematics extends the application scope of mathematics from precision to the field of fuzzy phenomena, and puts forward the theory of membership function, which largely determines whether things belong to concepts or not, so it is more reasonable to describe fuzzy problems than precise mathematics.

Similarly, the credit rating of enterprises is also a vague problem. It is difficult to judge its credit status with the concept of "yes" or "no" in precise mathematics. Therefore, it is more scientific to use fuzzy analysis method to comprehensively evaluate the credit status.

However, there are still doubts about the legitimacy of fuzzy mathematics in academic circles, because: first, fuzzy logic lacks learning ability and its application is limited. Secondly, it is difficult to guarantee the stability of fuzzy systems in theory. Secondly, fuzzy logic is not based on traditional mathematics, so it is difficult to verify the correctness of this logic system.

Enterprise credit rating method

Enterprise credit rating method, through studying the international advanced enterprise credit rating theory, enterprise credit rating thought, enterprise credit rating model and enterprise credit rating method, after years of continuous exploration, research, practice, innovation and accumulation, finally summed up a set of credit rating system that conforms to China's economic environment and is suitable for small and medium-sized enterprises in China. And established a complete enterprise credit rating database, industry-leading enterprise credit rating model and scientific enterprise credit rating method. Through the study of international advanced enterprise credit rating theory, enterprise credit rating thought, enterprise credit rating model and enterprise credit rating method, after years of continuous exploration, research, practice, innovation and accumulation, a set of credit rating system suitable for small and medium-sized enterprises in China is finally summarized, and a complete enterprise credit rating database, industry-leading enterprise credit rating model and. Enterprise credit network fully grasps the development pulse of various industries by using modern internet information technology, thus ensuring the accuracy and timeliness of enterprise credit rating results and improving the technical level of China's enterprise credit rating industry.

There are mainly these kinds of rating methods for enterprise credit rating, and domestic rating agencies generally adopt the last one.