Credit decision method

Abstract

The objective of this invention is to provide a credit judgment method that improves the accuracy of credit for companies that have the potential ability to repay by taking into account not only basic features of company information but also auxiliary features including diversity information. [Solution] The credit judgment method of the present invention is characterized by comprising the steps of: the computer acquiring, as input elements for calculating the default probability, at least a plurality of basic features as company information and a plurality of auxiliary features as company diversity information, input by an operator; and the computer multiplying each of the basic features and auxiliary features by a regression coefficient to calculate the default probability using inverse logit transformation.

Description

[Technical Field]

[0001] The present invention relates to a credit decision method for assisting in making a credit decision for a customer. [Background technology]

[0002] Patent Document 1 discloses an invention relating to a credit system that aims to provide a highly accurate credit system. The credit system of Patent Document 1 determines whether or not to start a transaction with a company analyzed by a financial information-utilizing credit analysis device. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2023-101969 Summary of the Invention [Problem to be solved by the invention]

[0004] As described above, according to Patent Document 1, since the credit model uses a financial information-based credit analysis device, small-scale companies with limited financial resources cannot receive sufficient credit transactions.

[0005] However, even among such companies, there are some that have potentially high repayment capacity, and conventional credit models that make credit decisions based only on basic company information have not been able to provide credit to these companies.

[0006] The present invention has been made in consideration of the above points, and aims to provide a credit judgment method that improves the accuracy of credit for companies that have the potential ability to repay by taking into account not only basic features of company information but also auxiliary features including diversity information. [Means for solving the problem]

[0007] One embodiment of the credit judgment method of the present invention is Small and medium-sized enterprises, start-ups, or venture companies A credit decision method executed by a computer to support a credit decision of the above, wherein the computer acquires, based on information input by an operator, at least a plurality of basic features as company information and a plurality of auxiliary features as company diversity information, as input elements for calculating a default probability. 1st step, the computer multiplies each basic feature amount and each auxiliary feature amount by a regression coefficient and calculates the default probability by inverse logit transformation. Third Step, Between the first step and the third step, there is provided a second step of determining whether or not necessary basic features and necessary auxiliary features have been acquired, and if all necessary basic features and necessary auxiliary features have been acquired in the second step, the process proceeds to the third step, or for those basic features and auxiliary features that cannot be acquired among the necessary basic features and necessary auxiliary features, specific features are input and the process proceeds to the third step. , characterized by: [Effects of the Invention]

[0008] According to the credit judgment method of the present invention, it is possible to improve the accuracy of credit judgment for companies that have the potential ability to repay, and to make judgments quickly. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a block diagram of a credit decision system (information processing device) according to this embodiment. [Figure 2] FIG. 2 is a flowchart of the credit judgment method according to this embodiment. [Figure 3] FIG. 3 is a table showing an example of feature quantities and regression coefficients for calculating the default probability. [Figure 4] FIG. 4 is a step diagram showing the flow of acquiring feature amounts. DETAILED DESCRIPTION OF THE INVENTION

[0010] The following describes in detail an embodiment of the present invention, but the following description is an example (representative example) of the embodiment of the present description, and the present invention is not limited to these contents as long as it does not go beyond the gist of the present invention.

[0011] <Background to the credit judgment method of this embodiment> The current credit model bases credit decisions on a company's current financial situation. In other words, it calculates default rates based on past performance. As a result, small and medium-sized enterprises with limited financial resources are given low credit ratings and are unable to receive sufficient credit transactions.

[0012] Incidentally, there are many excellent small and medium-sized enterprises, emerging companies, and venture companies that are expected to grow in the future, even if they currently have low financial capabilities. If such companies receive sufficient credit, their business will expand, which will ultimately lead to the development of Japan's industry. However, under the current credit model, potentially promising companies are unable to receive sufficient credit, which is an obstacle to business expansion.

[0013] Therefore, as a result of intensive research, the inventors have developed a credit judgment method that focuses on whether there are factors that can promote organizational growth and improve competitiveness, rather than current or short-term evaluation, and that makes it possible to provide credit transactions to companies that are expected to have high repayment capacity latent or in the future.

[0014] That is, the credit judgment method in this embodiment is characterized by improving the accuracy of credit for companies that have the potential ability to repay by taking into account not only basic (qualitative) features of company information but also auxiliary features including diversity information when calculating the default probability as a predicted value indicating the possibility of default in the future.

[0015] <Credit decision system (credit decision device) according to this embodiment> Figure 1 is a block diagram of a credit judgment system (information processing device) according to this embodiment. The credit judgment system 1 shown in Figure 1 comprises an input unit 3, a calculation unit 4, a storage unit 5, and a display unit 6. Note that although the credit judgment system 1 is divided into multiple functional units, this is for ease of explanation, and the functional units are not strictly separated, and the multiple functional units may operate as a single integrated functional unit.

[0016] The input unit 3 inputs the feature amounts necessary for calculating the default probability. The input values ​​of the feature amounts are, for example, "0: No, 1: Yes."

[0017] The calculation unit 4 calculates the default probability based on the feature amount input from the input unit 3. The default probability is the probability of falling into a default (non-performance of debt) state, expressed as a percentage (%).

[0018] The memory unit 5 stores the feature values ​​and the regression coefficients a input from the input unit 3. The regression coefficients can also be called likelihood ratios. The regression coefficients can be updated periodically or randomly by re-learning that takes into account economic conditions, etc. The larger the positive regression coefficient, the higher the probability of default, and the larger the negative coefficient, the lower the probability of default. The display unit 6 displays the default probability calculated by the calculation unit 4, information on the company that has been judged as being creditworthy, and the like on the screen. The input unit 3 inputs a plurality of basic features as company information and a plurality of auxiliary features as diversity information of the company.

[0019] The company information includes at least one of the following: governance information that can be read from the company register, company size information that can be read from the company website, and growth potential information that can be read from the company register or company public relations information on the company website. However, the company information is not limited to these.

[0020] Governance information includes at least one of the following: number of directors, whether the company is an owner-managed company, length of service as a director, number of director departures, location, whether the company is family-run, whether it has an auditor or accounting auditor, and whether it has a board of directors. However, governance information is not limited to these. Governance information can also be obtained from, for example, a system for searching for businesses covered by employee pension insurance and health insurance. The company size information includes at least one of the number of full-time employees, the number of stores, and the capital amount, but is not limited to these.

[0021] The growth information includes at least one of the following: the status of issuance of preferred stock / SO, frequency of social media posts, growth rate of number of employees and / or number of stores, and frequency of press releases. However, the growth information is not limited to these.

[0022] Other information available includes whether or not the company has made an official announcement, target values ​​that can be read from IR reports, whether or not it has engaged in ESG activities, whether or not it has engaged in CSR activities, business segments, the status of new business development, ROE, compliance information (whether or not it is an anti-social force, history of violations of laws and regulations, bad reputation, etc.), social insurance office information (whether or not it has had a health check, etc.), and details of its receivables (infrastructure / rent, entertainment, purchasing, loan repayments, capital / development investments, commission fees, etc.).

[0023] Diversity information includes at least one of the following: the ratio of female directors, the ratio of male to female employees, the ratio of nationalities other than Japanese, the age ratio, the ratio of people with disabilities employed, the ratio of female managers, whether or not telecommuting is used, and whether or not a childcare leave system is in place. However, diversity information is not limited to these.

[0024] Diversity information can be found, for example, from a company's website, a copy of the company register, social media, etc. For example, the "percentage of female directors" can be calculated from the names listed on the company's website.

[0025] The information that can be read from the website includes not only the company's own website but also the websites of other companies. For example, a job-hunting website can be an example of such a website.

[0026] 1 , an operator can input, via the input unit 7 of the credit decision system 1, a plurality of basic features as company information and information on which a plurality of auxiliary features as company diversity information are based. Alternatively, the operator can directly input the basic features and auxiliary features. That is, the operator can access the credit decision system 1 from a terminal device that can communicate with the credit decision system 1, and input, from the terminal device, a plurality of basic features as company information and a plurality of auxiliary features as company diversity information.

[0027] Furthermore, the credit decision result of the credit decision system 1 can be confirmed on the contract terminal device 2. The contract terminal device 2 is assumed to be a terminal device of a credit institution such as a financial institution or a credit company.

[0028] <Credit Judgment Method in the Present Embodiment> Next, the credit judgment method according to this embodiment will be described with reference to the flowchart of Fig. 2. Fig. 3 is a table showing an example of feature quantities and regression coefficients for calculating the default probability.

[0029] In step ST1 of Fig. 2, a plurality of basic feature amounts are acquired as company information. For example, a numerical value of 0 or 1 can be input, such as "0: No, 1: Yes" as shown in Fig. 3. The operator may input the basic feature quantities, or the computer may acquire the feature quantities based on the information input by the operator. In other words, when the operator selects the appropriate answer from multiple options displayed on the screen, the credit decision system 1 can automatically convert it into a feature quantity.

[0030] Although not limited to company information, it is preferable to include at least one of the following: governance information that can be read from the company register, company size information that can be read from the company website, and growth information that can be read from the company register or company public relations on the company website.

[0031] For example, as shown in Figure 3, there are input fields such as "Has there been an official gazette announcement?", "Has the company been established for more than 10 years?", "Has it conducted multiple businesses?", "Has it had capital of more than 100 million yen?", "Has it had a shareholder register administrator?", "Is it family-run?", "Does it issue stock options?", and "Has it established a board of directors?", and for each input field, enter "0: No, 1: Yes."

[0032] In step ST2 shown in FIG. 2, the credit judgment system 1 determines whether all necessary basic feature quantities have been obtained. If all have been obtained, the process proceeds to step ST3. If not, the process returns to step ST1. However, if there are items that cannot be obtained immediately, the acquisition can be postponed and the process can proceed to step ST3.

[0033] Also, the basic feature quantities necessary for credit judgment can be arbitrarily determined, and can be reduced or increased. The necessary basic feature quantities can determine specific input items, or can be the number of input items. That is, for example, “Is there an official gazette announcement?”, “Is it more than 10 years since establishment?”, “Is the capital more than 100 million yen?”, “Is there a shareholder register administrator?”, “Is stock option issued?”, “Is a board of directors installed?” described in FIG. 3 can be set as essential input items, and “Are there multiple businesses?”, “Is it family business?” can be set as optional input items. Control is performed so that the process cannot proceed to step ST3 if the essential input items have not been input. Or, although there are 8 input items described in FIG. 3, for example, input of 5 or more items can be made essential and controlled by a number.

[0034] Also, for example, when all 8 items shown in FIG. 3 are used as essential basic feature quantities, if there is an item that cannot be input anyway, it may be possible to select “0: No” or, in addition to “0: No, 1: Yes”, select “Unknown”. At this time, “Unknown” is set within the range of “0 < a < 1”. For example, according to the regression coefficient, the input value of the feature quantity of the input item can be given a width.

[0035] In step ST3 of FIG. 3, a plurality of auxiliary feature quantities as the diversity information of the company are obtained.

[0036] Although not limiting the diversity information, it preferably includes at least any one of the female ratio of directors, the male-female ratio of employees, the nationality ratio other than Japanese, the age ratio, and the employment ratio of disabled persons, the female ratio of management positions, the presence or absence of utilization of telecommuting, and the presence or absence of a parental leave system.

[0037] For example, as shown in Figure 3, there are input fields such as "Is the female ratio over 25%?", "Does the company have a work-from-home system and is it being utilized?", and "Does the company have a parental leave system and is it being utilized?", and you enter "0: No, 1: Yes" for each input field.

[0038] As shown in Figure 4, the input screen has input fields such as "Number of directors," "Number of female directors," "Years since establishment," "Whether the company is family-run," "Number of employees," and "Number of female employees," and the operator enters specific numerical values ​​into each input field.

[0039] On the computer side, as back-end processing, the calculation unit 4 shown in Figure 1 calculates whether "the ratio of female directors exceeds 25%" based on the input "number of directors" and "number of female directors," and if it is 25% or more, obtains a feature value of "1: Yes," and if it is 25% or less, obtains a feature value of "0: No."

[0040] Furthermore, on the computer side, as back-end processing, the calculation unit 4 shown in Figure 1 acquires a feature value of "1: Yes" based on the input "Years since establishment" if the company has been established for more than 10 years, and acquires a feature value of "0: No" if the company has been established for less than 10 years.

[0041] Furthermore, on the computer side, as back-end processing, the calculation unit 4 shown in FIG. 1 calculates whether "the female employee ratio exceeds 25%" based on the input "number of employees" and "number of female employees," and if it is 25% or more, a feature value of "1: yes" is acquired, and if it is 25% or less, a feature value of "0: no" is acquired.

[0042] However, in emerging companies and venture businesses, there are many items that do not correspond to the input items of basic features or are unknown, making it difficult to make accurate credit decisions. For example, questions such as "Established for more than 10 years?" and "Capital of more than 100 million yen" shown in Figure 3 are unlikely to apply, and the input value for the feature is "0: No."

[0043] Startups and venture companies do not have sufficient financial performance. For this reason, it is difficult to evaluate their future potential using traditional credit assessment methods. It is impossible to judge the viability of business continuity or business plans based on financial statements whose reliability is uncertain.

[0044] Therefore, auxiliary features as diversity information are set as input items, and the possibility of improving the long-term value of the entire organization is evaluated by focusing on whether there are elements that can promote organizational growth and improve competitiveness. Furthermore, in this embodiment, the default probability is calculated taking into account not only quantitative information but also qualitative information.

[0045] 2, the credit decision system 1 determines whether all necessary diversity information has been input. If all necessary diversity information has been acquired, the process proceeds to step ST5. If there is insufficient diversity information, the process returns to step ST3.

[0046] The auxiliary features required for credit decisions can be determined arbitrarily and can be increased or decreased. The necessary auxiliary features can be determined as specific input items or can be the number of input items. For example, among the ratio of female directors, the ratio of male to female employees, the ratio of nationalities other than Japanese, the age ratio, the ratio of people with disabilities hired, the ratio of female managers, whether or not work-from-home is used, and whether or not a parental leave system is available, the ratio of female directors, the ratio of male to female employees, the age ratio, the ratio of female managers, whether or not work-from-home is used, and whether or not a parental leave system is available can be set as essential diversity information, while the ratio of nationalities other than Japanese and the ratio of people with disabilities hired can be set as optional diversity information.

[0047] It may be difficult to obtain diversity information from a company's website or registered copy of its corporate register, etc. In such cases, it is possible to obtain information from other companies' websites, such as job-hunting sites, or from social media.

[0048] In addition, if there is an item that cannot be input into the essential diversity information, you may select "0: No" or, in addition to "0: No, 1: Yes", you may also be able to select "Unknown". At this time, "Unknown" is within the range of "0 < a < 1". For example, according to the regression coefficient, the input value of the feature amount of the input item can have a width.

[0049] Since the importance of diversity information changes depending on the era, industry type, etc., it is desirable to periodically review (re-learn) input items and regression coefficients, etc., to improve the credit judgment accuracy by the credit judgment system 1.

[0050] And in step ST5 shown in FIG. 2, the default probability is calculated. The regression coefficient a required to calculate the default probability is stored in advance in the storage unit of FIG. 1. The regression coefficient is a predetermined numerical value and indicates the accuracy of the default prediction value. It is calculated by machine learning based on a large number of corporate data. In the calculation unit 4, using the input feature amount and the regression coefficient associated with each feature amount, the default probability is calculated based on the following (Equation 1).

[0051] inv_logit(a1x1+a2x2+···a , , ,

[0053] ,

[0052] , n , x n +b) (Equation 1) (Here, a1, a2…a n (n is an integer of 1 or more) indicates the regression coefficient, x1, x2…x n (n is an integer of 1 or more) indicates the feature amount, and b indicates the intercept).

[0052] (Equation 1) is expressed as the probability obtained by taking the inverse logit of the feature amount (input value) × regression coefficient + intercept. Note that the intercept b is used auxiliaryly to improve the accuracy of the default prediction value and is set individually according to each credit diagnosis model.

[0053] In the case of a conventional credit review that mainly focuses on financial matters, the market average default probability is 0.9 to 1.1%, whereas in the case of a credit review using the method of this embodiment that uses diversity information and qualitative information, the default probability is 0.09%, which is approximately one-tenth of the previous credit review.

[0054] As a result, even companies that would not have passed conventional credit screening can pass the screening and receive speedy credit transactions by utilizing the credit screening of this embodiment. [Explanation of symbols]

[0055] 1: Credit decision system 2: Contract terminal device 3: Input section 4: Arithmetic section 5: Storage section 6:Display section 7: Input section

Claims

1. A credit decision-making method executed by a computer to support credit decisions for small and medium-sized enterprises, emerging companies, or venture companies as customers, comprising: a first step in which the computer acquires, based on information input by an operator, at least a plurality of basic features as company information and a plurality of auxiliary features as company diversity information, as input elements for calculating a default probability; a third step in which the computer multiplies each of the basic feature amounts and each of the auxiliary feature amounts by a regression coefficient to calculate the default probability by inverse logit transformation; a second step between the first step and the third step for determining whether necessary basic features and necessary auxiliary features have been acquired, and if all of the necessary basic features and necessary auxiliary features have been acquired in the second step, the method proceeds to the third step, or for those basic features and auxiliary features that cannot be acquired among the necessary basic features and necessary auxiliary features, a specific feature is input and the method proceeds to the third step; A credit decision method comprising:

2. A credit judgment method as described in claim 1, wherein the basic feature and the auxiliary feature are 0 or 1, and the specific feature is 0 or 1 or a value greater than 0 and less than 1.

3. A credit judgment method as described in claim 1, in which the basic features and auxiliary features are obtained through back-end processing of the computer.

4. 2. The credit judgment method of claim 1, wherein the diversity information includes at least one of the following: the ratio of female directors, the ratio of male to female employees, the ratio of nationalities other than Japanese, the age ratio, the ratio of people with disabilities hired, the ratio of female managers, whether or not work-from-home arrangements are used, and whether or not a parental leave system is in place.

5. The credit judgment method of claim 1, wherein the company information includes at least one of governance information that can be read from a copy of the company register, company size information that can be read from a website, and growth potential information that can be read from a copy of the company register or company public relations on a website.

6. The credit judgment method of claim 5, wherein the governance information includes at least one of the number of directors, whether the company is an owner-managed company, the length of time that directors have served, the number of directors who have left office, the location, whether the company is a family-run business, whether there is an auditor or an accounting auditor, and whether there is a board of directors.

7. The credit judgment method according to claim 5 , wherein the company size information includes at least one of the number of full-time employees, the number of stores, and the capital amount.

8. The credit judgment method according to claim 5, wherein the growth information includes at least one of the following: preferred stock / SO issuance status, frequency of SNS posts, growth rate of number of employees and / or number of stores, and frequency of press releases.

9. 2. The credit judgment method according to claim 1, wherein the default probability is expressed by the following formula 1: (Equation 1) inv_logit(a 1 x 1 + a 2 x 2 + ··· a n x n + b) (Equation 1) (where a 1 , a 2 …a n (n is an integer of 1 or more) indicates the regression coefficient, and x 1 , x 2 …x n (n is an integer of 1 or more) indicates a feature amount, and b indicates an intercept).

Images