Optimal credit strategy for MSMEs

. We address two main issues: one is to quantify the credit risk of enterprises and establish a complete credit risk system, and the other is to give the optimal credit risk strategy for banks. We first analyze and pre-process the data. From the data, we extracted a series of indicators such as the total amount of input and output, and the length of operation. We analyze the credit risk in three directions: strength, stability of supply and demand, and creditworthiness, and establish a credit risk quantification system for the enterprise. Then we quantify the credit risk of the enterprise by entropy method and TOPSIS. Second, a function is fitted to the bank's customer churn rate and the bank's lending rate. Using a planning-type model, the credit decisions of the firms are required to be given. We follow the principle of maximizing benefits and minimizing risks to build a multi-objective planning model. We base on the scores of each firm that have been solved, for the classification of firms, and follow the principle of low credit risk, low lending rate, and set the corresponding lending rate for each type of firm. The model is solved by the through-order solution method, and the linear weighting method is used to test the comparison. The credit decision for each enterprise is given.


Introduction
In practice, since MSMEs are relatively small in scale and also lack collateral assets, banks usually provide loans to enterprises with strong strength and stable supply and demand relationships based on credit policies, information on their trading notes and the influence of upstream and downstream enterprises, and can offer preferential interest rates to enterprises with high creditworthiness and low credit risks. Banks first assess the credit risk of MSMEs based on their strength and reputation, and then determine whether to lend and credit strategies such as loan amount, interest rate and maturity based on credit risk and other factors. This requires a reasonable and efficient credit risk quantification system and an optimal credit decision model based on credit risk. The study of the optimal credit decision of banks for MSMEs is beneficial to the healthy development of banks and the stability of the financial market on the one hand; on the other hand, it is beneficial to increase the chances of obtaining loans for MSMEs and promote the development of MSMEs.

Approaches
Firstly, for data pre-processing, enterprises with poor creditworthiness, i.e., those with default records and creditworthiness registration of D, were excluded. Then, evaluation indicators were determined, and through references, nine indicators such as total amount of input, standard deviation of monthly input invoices, and creditworthiness score were identified for evaluating credit risk. And the TOPSIS method based on entropy weight method is used to quantify the credit risk of each enterprise by totaling and scoring each enterprise. [1] A function fit was performed by MATLAB for customer churn rate and annual interest rate, and by trying different fitting functions, a function curve with less error and more accuracy was determined. Finally, we follow the principle of maximizing benefits and minimizing risks to build a multi-objective planning model, which we base on the scores of each enterprise. Following the principle of low credit risk, the loan interest rate is low, and the corresponding loan interest rate is set for each type of enterprise. The model is solved by the through-order solution method and is tested for comparison using a linear weighting method.

Data pre-processing
According to the realistic requirements, banks in principle do not lend to enterprises with default records and credit ratings of D. Therefore, these enterprises are excluded from the data pre-processing, and 96 enterprises with no default records and credit ratings of D or above are screened out, and only these enterprises are considered in the subsequent model solving and credit strategy.

Determination of evaluation indicators
Banks tend to lend to enterprises with strong and stable supply and demand relationships, in addition to the ability to repay loans and the degree of creditworthiness are also important factors affecting the credit strategy of banks. In this paper, nine indicators are constructed to evaluate the credit risk of enterprises in terms of their strength, business stability and creditworthiness, such as total amount of input, standard deviation of monthly input invoices and credit score. The evaluation system is schematized as follows. [1]  After determining each scientifically valid evaluation index, the importance evaluation model is constructed, and the weight of each index is calculated by the entropy weighting method to establish a TOPSIS-based multiindicator evaluation model, the higher the score can reflect the importance of the supplier, and the larger the value, the higher the degree of importance. Entropy originally comes from thermodynamics, then introduced by Shannon into information theory, according to the definition and principle of entropy, when the system may be in several different states, the probability of each state is , then the entropy of the system can be defined as Entropy weighting method to solve the weighting steps: where i L denotes the customer churn rate when the bank lends to the ith firm, and i I denotes the annual interest rate of the loan for the ith firm. For this problem, this paper writes a program to solve it using Matlab software with maximizing bank profit as the priority objective and minimizing loss as the secondary objective. [4,5]

Result
The combined scores of the 96 companies are shown below. The basic idea of the priority method for solving multiobjective planning is to divide the objectives into different priority levels according to their importance, and first find the optimal value of the objective with high priority, and then find the objective function with low priority under the condition that the objective with high priority gets no less than the optimal value. The final results are presented in part as follows.