JP2020013229A - Device, method and program for calculating default probability - Google Patents

Abstract

To directly calculate a default probability for a given desired loan amount in a method of calculating a default probability of a borrower.SOLUTION: A device 100 receives a desired loan amount from a company terminal 110 (S201). Next, the device 100 acquires monthly cash balances of M past months (M is an integer equal to or larger than 1) from a database 120 in which past cash balances are stored in association with a borrower company using the company terminal 110 or its company ID (S202). The device 100 predicts a cash balance distribution of the company after N months (N is an integer equal to or larger than 1) on the basis of the cash balances of M past months (S203). After acquiring the cash balance distribution after N months, the device 100 calculates a probability that the cash balance may be equal to or smaller than the desired loan amount, as a default probability (S204). Then, an interest rate consistent with the calculated default probability is determined as needed (S205), and loan information including the interest rate is transmitted to the company terminal 110 (S206).SELECTED DRAWING: Figure 2

Description

  The present invention relates to an apparatus, a method, and a program for calculating a probability of default.

  In financial institutions, as a screening method for business financing, a scoring model is being used in addition to the conventional method in which a screening person analyzes and determines a financial statement, a trial balance, and the like. The scoring model scores the creditworthiness of a potential borrower, and performs a rating, thereby enabling efficient examination of the financing.

  A typical example of the scoring model is a statistical model. As explanatory variables of the model, indicators such as the receivables turnover period, the equity ratio, and the quick ratio are selected from the financial statements before the loan. Then, a delay in payment of principal and interest one year later is defined as “default”, and the presence or absence of a default after financing is observed in a financial statement after financing. By performing logistic regression analysis using the default probability as an objective variable and one or more selected financial indices as an explanatory variable, a scoring model can be statistically constructed. For a new borrower, a predicted value of the default probability can be obtained as a score from the past financial statements of the company.

  Each financial institution builds such a model based on its own loan performance data, and ranks potential lenders in consideration of qualitative factors such as manager's ability and qualities that do not appear in the financial statements. If the rating is below a certain level, financing is not possible.If the rating is higher than that, financing is possible.If financing is possible, the financing conditions are set within the financing limit set according to the rating. Judgment such as judgment is being performed.

  However, the conventional scoring model for business financing is to judge the possibility of financing based on the rating of the borrower based on the probability of default in one year, and to directly evaluate the risk for each loan Not something. This creates a loss of financing opportunities, such as cases where specific individual financing is possible but the rating is low from the most recent financial statement, but the existing scoring model There is a need for a new model that replaces or complements.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide a device, a method, and a program for calculating the probability of default of a borrower, the probability of default for a given desired loan amount. Is to calculate directly.

  In order to achieve such an object, a first aspect of the present invention is a method of calculating a probability of default with respect to a desired amount of financing, the method being provided after N months (N is an integer of 1 or more) of a borrowed company. Acquiring a cash balance distribution; and calculating a probability that the cash balance is equal to or less than the desired loan amount in the cash balance distribution as the default probability.

  In a second aspect of the present invention, in the first aspect, N is 3 or less.

  Further, a third aspect of the present invention, in the first or second aspect, is characterized in that the financing of the desired financing amount is financing on condition of lump-sum repayment.

  In a fourth aspect of the present invention, in any one of the first to third aspects, the N, which is the desired loan amount or the loan period, is received from a company terminal of the borrower company. I do.

  According to a fifth aspect of the present invention, in any one of the first to third aspects, the desired loan amount or the loan period N is given as an initial value.

  Further, according to a sixth aspect of the present invention, in any one of the first to fifth aspects, an interest rate or availability of a loan to the borrower is determined according to the probability of default. .

  Further, a seventh aspect of the present invention, in the sixth aspect, further includes a step of transmitting financing information including the interest rate or the propriety to a corporate terminal of the borrower company.

  In an eighth aspect of the present invention, based on the seventh aspect, a step of receiving the change instruction of the loan information from the company terminal, and the step of transmitting the loan information including the changed interest rate or interest based on the change instruction. Retransmitting to the enterprise terminal.

  In a ninth aspect of the present invention, in any one of the first to eighth aspects, the cash balance distribution is generated by Monte Carlo simulation.

  In a tenth aspect of the present invention, in any one of the first to ninth aspects, the cash balance distribution is generated by machine learning.

  An eleventh aspect of the present invention is the ninth or tenth aspect, wherein the cash balance distribution is one or a plurality of webs representing a deposit and withdrawal history of one or a plurality of financial institutions acquired via a computer network. It is characterized by being generated using detailed data.

  According to a twelfth aspect of the present invention, in any one of the ninth to eleventh aspects, the cash balance distribution is stored in advance in association with the borrower company.

  In a thirteenth aspect of the present invention, in any one of the first to twelfth aspects, the calculation is performed in an account service of the borrower company in an accounting service provided on a server for cloud computing. Is executed by the server in response to a change in the accounting data linked to.

  A fourteenth aspect of the present invention is a program for causing a computer to execute a method of calculating a probability of default with respect to a desired amount of loan, the method comprising: after N months (where N is 1) A step of obtaining a cash balance distribution of the above integer) and a step of calculating a probability that the cash balance in the cash balance distribution is equal to or less than the desired loan amount as the default probability.

  Further, a fifteenth aspect of the present invention is an apparatus for calculating a probability of default with respect to a desired loan amount, and obtains a cash balance distribution after N months (N is an integer of 1 or more) of a borrowed company, The method includes calculating a probability that the cash balance is equal to or less than the desired loan amount in the cash balance distribution as the default probability.

  According to one aspect of the present invention, by referring to the cash balance distribution of the borrower company after N months, the probability that the cash balance is equal to or less than the desired loan amount in the cash balance distribution is calculated as the probability of default. In addition, it becomes possible to directly determine individual loan conditions for a given desired loan amount.

FIG. 1 is a diagram illustrating an apparatus according to a first embodiment of the present invention. FIG. 3 is a diagram illustrating a method according to the first embodiment of the present invention. It is a figure showing the prediction method of cash balance distribution concerning a 2nd embodiment of the present invention. It is a figure showing the example of cash balance change concerning a 2nd embodiment of the present invention. It is a figure showing the example of the cash balance difference concerning a 2nd embodiment of the present invention. It is a figure showing an example of cash balance distribution concerning a 2nd embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

(First embodiment)
FIG. 1 shows an apparatus according to a first embodiment of the present invention. The device 100 includes a communication unit 101 such as a communication interface, a processing unit 102 such as a processor and a CPU, and a storage unit 103 including a storage device such as a memory and a hard disk or a storage medium. By executing the functions, the functions described below can be realized. The device 100 may include one or a plurality of devices or servers, and can be configured using resources on the cloud. The program may include one or more programs, and may be recorded on a computer-readable storage medium to be a non-transitory program product.

  First, the device 100 receives a desired loan amount such as 1,000,000 yen from the company terminal 110 via a computer network such as the Internet (S201). The financing period may be received from the company terminal 110. For example, a web page for inputting a desired financing amount is displayed on the company terminal 110, and the web page or the financing displayed on the web page is displayed. The product may be stored in the storage device or storage medium 120 accessible from the device 100 or the device 100 in association with the loan period of the loan for which the desired loan amount is input on the web page. In addition, the loan amount or its width can be associated with the loan period.

  Next, the database 120 stores the past cash balance in association with the borrower company using the company terminal 110 or the company ID thereof, and the apparatus 100 stores the past M months (M is an integer of 1 or more) for the past M months. The cash balance for each month is acquired (S202). The balance may be, for example, a month-end balance. The cash balance of each company can be stored by acquiring web statement data representing the deposit and withdrawal history of a financial institution via a computer network, and together with financing conditions such as a desired loan amount from the company terminal 110. Alternatively, it can be received separately from the loan conditions. Here, M is preferably 12 or more.

  The device 100 predicts the cash balance distribution of the company after N months (N is an integer of 1 or more) based on the cash balance for the past M months (S203). Details of the prediction will be described later. N month corresponds to the loan period, and “after N month” may be more specifically defined as the end of the last month from the last month. More generally, when the cash balance is the balance on the reference date including days other than the end of the month, “after N months” is the corresponding date N months after the latest reference date or the date to which the corresponding day belongs. May be the end of the month. It is also conceivable that “after N months” is the corresponding day N months after the scheduled loan date or the end of the month to which the corresponding day belongs. Here, N is preferably 12 or less, more preferably 3 or less.

  After the cash balance distribution after N months is obtained, the apparatus 100 calculates the probability that the cash balance is equal to or less than the desired loan amount in the cash balance distribution as the default probability (S204). Considering lump-sum repayment of the loan amount, if the cash balance N months after the latest base date or scheduled loan date is greater than or equal to the desired loan amount, the entire repayment is possible on the repayment date, and it will be less than that If so, it can be said that payment is likely to be delayed. Even if the cash balance is less than or less than the desired loan amount, if the difference from the amount is within a predetermined range, it is considered that there is a high possibility that the debt can be fulfilled, and the probability of default may be calculated. Further, instead of the probability of default, the repayment probability of repayment may be calculated.

  Then, the interest rate according to the probability of default or repayment calculated as necessary is determined (S205), and the loan information including the interest rate is transmitted to the enterprise terminal 110 and displayed on the display screen. The loan information includes the above interest rate, and may include a repayment term or a loan period, a loan amount, and the like (S206).

  The loan information may include whether or not a loan can be made. For example, if the interest rate calculated based on the probability of default or repayment exceeds the upper limit interest rate, it may be determined that the loan cannot be made.

  In the present embodiment, the probability of default is calculated by predicting the repayment date or the distribution of the cash balance immediately before or before the repayment date based on the change in the cash balance of the borrower for the loan amount requested in this way. This makes it possible to directly determine individual loan conditions.

  In particular, by using a monthly index of cash balance in each month instead of a year as indices obtained from financial statements, make loan decisions according to the current situation of each company that does not appear in financial statements Can be. Conventional scoring models are often not suitable for short-term loans with a repayment period of one year or less, and ultra-short-term loans with a repayment period of less than three months.

  In addition, although it is not easy for a financial statement used in the conventional scoring model to see dressing and fake, it is difficult to find out the cash balance of each company in one or a plurality of web accounts of accounts in which the company has one or more financial institutions. When acquiring based on data, such fraud can be eliminated or suppressed, and the accuracy of the probability of default can be increased.

  The calculated estimated cash balance distribution after N months is stored in the database 120 or the storage unit 103 of the device 100 in association with the borrower company or its company ID, and can be accessed from the processing unit 102 as necessary. It can be. By calculating in advance various values of N, it is possible to quickly make a loan decision under various conditions. Also, the forecasted cash balance distribution can be regenerated and updated on the 5th, 10th, 15th, 20th, 25th, 30th, the end of the month, the beginning of the month, the end of the month, and the like every month.

  In the present embodiment, it is assumed that the apparatus 100 receives the desired loan amount from the company terminal 110. However, an initial value is set for each company or each industry, or an initial value common to the entire company is set. Or give some initial value.

  Also, an example in which data for the last M months is used as the past cash balance has been described. However, it is sufficient to use not the latest data but data on a change in the cash balance that can represent the current status of the company. There is.

  In addition, unless there is a description such as “only based on xx”, “only according to xx”, or “only xx”, in this specification, additional information is used. Note that it is envisaged that this may also be considered.

  Also, just in case, any method, program, terminal, device, server or system (hereinafter referred to as “method or the like”) may perform an operation different from the operation described in this specification. The aspect is directed to the same operation as any of the operations described in the present specification, and the existence of an operation different from the operation described in the present specification means that the method and the like are described in each of the present invention. It is noted that the embodiment is not out of the scope of the embodiment.

(Second embodiment)
Prediction of the cash balance distribution of the company after N months based on the cash balance for the past M months using the Monte Carlo simulation will be described.

As an example, the value at the end of December for the following December is used as M as 12. Further, the loan period N is set to 3, and the cash balance distribution at the end of March of the following year is predicted. Attempts _{i 0} of Monte Carlo simulation is 10,000 times. FIG. 4 shows the transition of the cash balance.

First, a difference of (M-1) points is obtained from the cash balance for M months (S301). FIG. 5 shows the transition of the difference in the cash balance. Hereinafter, the cash balance for each month _{B m (1 ≦ m ≦ 12} ), the difference _{B m -B m-1} a (2 ≦ m ≦ 12) is expressed as .DELTA.B _m. As initial values, i is set to 1 and n is set to 0 (S302).

i th after n months in trial (1 ≦ n ≦ N) of the predicted cash balances _{b in,} the initial value is expressed as _{b i0} in i-th trial, the i satisfying 1 ≦ _i ≦ i ₀ (S303 ), N times .DELTA.B _m is financed months extracted randomly added to the initial value _{b i0} (S304 to S306). Here, the initial value b _i0 may be B ₁₂ is the most recent past cash balances, or may be a cash balance in a subsequent loan date, etc. date for. In this example, 2,281,734 yen was used as the loan balance. When the addition is completed N times, n is reset to 0, and i is incremented (S307). When you have finished i ₀ trials, because i ₀ one of the predicted cash balance b _iN is obtained, it is possible to generate or store the cash balance distribution after N month (S308). FIG. 6 shows a predicted cash balance distribution. In using the generated cash balance distribution, it is not always necessary to make the distribution diagram as shown in FIG.

  According to the method of the present embodiment, prediction of the future cash balance distribution based on only the past cash balance change of the borrower, and hence the default using the same, without the need for the loaned data for implementation. The probability can be calculated. In the above example, assuming that the desired loan amount is 1 million yen, the probability that the cash balance after March will be less than or equal to the desired loan amount is 0.029, that is, 2.9%.

  In the present embodiment, all the data for the past M months is used. However, a known pre-processing for the data for improving the prediction accuracy, such as an outlier processing, may be applied.

  In the above description, the Monte Carlo simulation is taken as an example, but it is possible to predict the cash balance or the cash balance distribution after N months by machine learning. Specifically, various methods such as SVM, decision tree, and deep learning can be used.

(Third embodiment)
In the first embodiment, it is assumed that a desired amount of financing is input from the company terminal 110 in some form, but it is not an independent financing service but a function of an accounting service. , Find a characteristic use of the predicted balance distribution.

  First, in the accounting service, in response to the active operation of the accounting data from the enterprise terminal 110, the desired loan amount and the loan period are set as initial values, the probability of default is calculated, and the interest rate is calculated. By determining loan conditions such as the above, loan information including the loan conditions can be transmitted to the company terminal 110 and displayed on the display screen. Examples of active operations include registering accounts payable or receivable, registering received invoices, issuing prepared invoices, registering payments, and the like.

  Second, when the accounting service is a service provided on a server for cloud computing, the accounting service is received from an external service associated with an account of a user using the enterprise terminal 110. Thus, it is possible to determine loan conditions. Examples of the accounting data include bill data, web statement data, and the like.

  Third, it is possible to determine a loan condition or the like in accordance with the fact that the accounting data in the account of the user using the company terminal 110 satisfies a predetermined condition. The prescribed conditions are that the accounts receivable is greater than or equal to a prescribed amount, the deposit balance is less than or equal to a prescribed amount, that the repayment of the borrowing has been completed or is scheduled to be completed, and that the withdrawal of the prescribed amount or more And the like.

  Fourth, if the accounting service is a service that allows an accounting firm, tax accountant office, or other advisor to which the user using the company terminal 110 is an advisor to view the accounting data of the user, the accounting data is audited. As a result, there are times when it is deemed appropriate to get a loan. In such a case, the advisor can add a comment to that effect in association with the accounting data, and upon the occurrence of the comment, determine a loan condition or the like. The advisor may check the obtained financing information and then enter a comment that can be viewed by the user together with the financing information.

  The financing information transmitted to the corporate terminal 110 as a result of the processing automatically performed in response to the change in the accounting data linked to the account of the borrower company is The desired amount or financing period can be variable. The apparatus 100 can receive the changed loan amount or the loan period as a change instruction, and can retransmit the loan information. In particular, a movable element such as a slider is displayed on the display screen of the corporate terminal 110, and the desired loan amount or the loan period changes according to the user moving the element, and under the changed conditions, A UI can be provided in which interest rates are redisplayed.

Reference Signs List 100 device 101 communication unit 102 processing unit 103 storage unit 110 company terminal 120 database

Claims (15)

A method of calculating a probability of default with respect to a desired loan amount,
Obtaining a cash balance distribution after N months (N is an integer of 1 or more) of the borrower company;
Calculating the probability that the cash balance in the cash balance distribution is less than or less than the desired loan amount as the default probability.   The method of claim 1, wherein N is 3 or less.   The method according to claim 1, wherein the loan for the desired loan amount is a loan contingent on lump-sum repayment.   The method according to any one of claims 1 to 3, wherein the N which is the loan amount or the loan period is received from a company terminal of the borrower company.   The method according to any one of claims 1 to 3, wherein the loan amount or the loan period (N) is provided as an initial value.   The method according to any one of claims 1 to 5, wherein the interest rate or availability of the loan to the borrower is determined according to the probability of default.   7. The method according to claim 6, further comprising transmitting the loan information including the interest rate or the propriety to a corporate terminal of the borrower company. Receiving an instruction to change the loan information from the enterprise terminal;
The method according to claim 7, further comprising retransmitting loan information including a changed interest rate or interest based on the change instruction to the enterprise terminal.   The method according to claim 1, wherein the cash balance distribution is generated by Monte Carlo simulation.   The method according to any of the preceding claims, wherein the cash balance distribution is generated by machine learning.   11. The cash balance distribution according to claim 9, wherein the cash balance distribution is generated using one or a plurality of web statement data representing a deposit and withdrawal history of one or a plurality of financial institutions acquired via a computer network. the method of.   The method according to any one of claims 9 to 11, wherein the cash balance distribution is stored in advance in association with the borrower company.   The calculation is executed by the server in response to a change in accounting data associated with the account of the borrower company in an accounting service provided on a server for cloud computing. A method according to any of claims 1 to 12, characterized in that: A program for even executing a method of calculating a probability of default with respect to a desired loan amount, the method comprising:
Obtaining a cash balance distribution after N months (N is an integer of 1 or more) of the borrower company;
Calculating a probability that the cash balance is equal to or less than the desired loan amount in the cash balance distribution as the default probability. An apparatus for calculating a probability of default with respect to a desired loan amount,
Obtain the cash balance distribution of the borrower company after N months (N is an integer of 1 or more),
An apparatus comprising calculating a probability that a cash balance in the cash balance distribution is equal to or less than the desired loan amount as the default probability.

Images