JP2017120547A - Commitment line additional withdrawal rate estimation model - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve several problems in which a current commitment line additional withdrawal estimation method relies on a small number of default results, therefor, (1) an estimation model might be unstable as an estimation value (for example, when the estimation model is annually updated, a deviation width of the estimation model might be larger), (2) a classification estimation based on a segment (financial instrument category, rating, industry sector and the like) is difficult, and (3) a benchmark model is required for checking adequacy of the current estimation method, and the like.SOLUTION: Based on a default destination + a rating tabulation by a rating with respect to debtors at a non-default low rating destination, a commitment line additional withdrawal rate is calculated that is higher in stability.SELECTED DRAWING: Figure 7

Description

  The present invention relates to a commitment line additional withdrawal rate estimation model. Specifically, the present invention relates to a computer, a method, and a computer program for estimating a more stable commitment line additional withdrawal rate by summarizing the withdrawal rate performance for a debtor whose rating has declined by rating.

  According to the FSA announcement (the Basel III rules are announced to domestic banks), financial institutions such as domestic banks operating internationally have achieved a certain level of indicators such as capital adequacy ratio and Tier 1 ratio. It is required to ensure (capital regulations). For example, the total capital ratio is calculated as: Equity / Risk asset x 100 (%), and financial institutions operating internationally are required to maintain at least 8%. This regulation aims to protect depositors by ensuring the soundness of financial institutions and avoiding bankruptcies.

  “Equity” in the total capital ratio is calculated as the sum of each index of common equity tier 1 + other tier 1 + tier 2, and “risk asset” is credit risk asset + (market risk equivalent + operational risk equivalent) x 12.5 These formulas and indicators are defined by the Basel Committee. Among these, “credit risk assets” can be calculated by EAD × risk weight. EAD is required to take into account future additional withdrawals related to the commitment line. The risk weight can be calculated according to a risk weight function defined by the Basel Committee using PD, LGD, and maturity as input variables. Each input parameter is estimated by each financial institution. A summary of each input parameter is shown in Table 1.

  Of these, the commitment line additional withdrawal rate is currently calculated as follows: “Additional withdrawal amount” obtained by subtracting the used amount one year ago from the currently used amount, as shown in FIG. It can be calculated by dividing by “.

  However, the current estimation method based on the commitment line additional withdrawal rate relies on a small number of default results, and (1) there is a possibility that the estimation value becomes unstable (for example, the estimation model is updated annually) (2) Difficult to classify by segment (product type, rating, type of industry, etc.), (3) A benchmark model to confirm the validity of the current estimation method is required There are some problems such as.

The present invention was made to achieve such an object, and an embodiment thereof is a computer that estimates a commitment line additional withdrawal rate in order to calculate a capital institution ratio of a financial institution, The computer
Obtain ratings of multiple obligors, the ratings are set by the financial institution in accordance with its own provisions, and include past and current ratings,
From the difference between the previous rating and the current rating, calculate a rating decline in a predetermined period from the past to the present,
On the basis of the rating reduction range, among the plurality of debtors, determine a debtor whose rating has declined in the predetermined period,
For the debtor whose rating has declined, calculate the additional withdrawal amount from the credit line in the predetermined period,
Based on the additional withdrawal amount, a commitment line additional withdrawal rate is calculated,
It is configured to calculate an average value of the additional withdrawal rate of the commitment line for each rating reduction range.

  Further, the computer according to the preceding paragraph formulates a relationship between the rating decline and the average value of the commitment line additional withdrawal rate by regression analysis, and the rating decline is expressed in the formulated relational expression. , The average value of the commitment line additional withdrawal rate is corrected, and the withdrawal rate estimated value is further estimated.

  Further, the computer described in the preceding paragraph is further configured to perform a weighted average of the estimated value of the withdrawal rate by the number of obligors by rating or the number of expected DFs, and the expected number of DFs becomes a default in the future. It is characterized by the number of debtors expected to end.

In the computer described in the previous paragraph,
The expected DF number is calculated by multiplying the number of obligors by rating and the PD value.

  Furthermore, in the computer described in the preceding two paragraphs, an estimated value weighted averaged by the number of obligors by rating or the expected number of DFs is used as an input parameter for capital ratio.

Another embodiment of the present invention is a computer-implemented method for estimating a commitment line additional withdrawal rate to calculate a capital institution ratio of a financial institution, the method comprising:
Obtaining a rating of a plurality of debtors, wherein the rating is set by the financial institution in accordance with its own provisions and includes a past rating and a current rating;
A step of calculating a rating decline in a predetermined period from the past to the present from a difference between the past rating and the current rating;
Determining a debtor whose rating has declined during the predetermined period among the plurality of debtors based on the rating decline width;
Calculating an additional withdrawal amount from the credit line in the predetermined period for the debtor whose rating has declined;
Calculating a commitment line additional withdrawal rate based on the additional withdrawal amount;
Calculating an average value of the additional withdrawal rate for each commitment line.

Furthermore, another embodiment of the present invention is a computer program for causing a computer to execute a computer-implemented method for estimating a commitment line additional withdrawal rate in order to calculate a capital institution ratio of a financial institution. When the program is executed by the computer, the computer
Obtaining a rating of a plurality of debtors, wherein the rating is set by the financial institution according to its own provisions, and includes a past rating and a current rating;
From the difference between the previous rating and the current rating, let us calculate the rating decline in the predetermined period from the past to the present,
Based on the rating reduction range, among the plurality of debtors, the debtor whose rating has declined in the predetermined period is determined,
Let the debtor whose rating declines calculate the additional withdrawal amount from the credit line in the predetermined period,
Based on the additional withdrawal amount, the commitment line additional withdrawal rate is calculated,
An average value of the additional withdrawal rate of the commitment line is calculated for each rating reduction range.

  As described above, according to the present invention, it is possible to calculate a more stable commitment line additional withdrawal rate (and hence an equity ratio) as an estimated value, enabling classification estimation by segment, and It can be used as a benchmark model to confirm the validity of the current estimation method.

It is a figure which shows the structure of the estimation server which concerns on one Embodiment of this invention. It is a figure which shows the outline | summary of the commitment line additional drawing-out rate which concerns on one Embodiment of this invention. It is a figure which shows the data stored in the drawer performance data storage part which concerns on one Embodiment of this invention. It is a figure which shows the data stored in the total data memory | storage part classified by rating fall width which concerns on one Embodiment of this invention. It is a figure which shows the data stored in the total data storage part classified by rating which concerns on one Embodiment of this invention. It is a figure which shows the data stored in the weighted average data storage part which concerns on one Embodiment of this invention. It is a flowchart which shows the commitment line additional withdrawing rate total process which concerns on one Embodiment of this invention. It is a flowchart which shows the detail of the commitment line additional drawing-out rate total process which concerns on one Embodiment of this invention.

  The configuration of the estimation server 100 according to the embodiment of the present invention will be described in detail. FIG. 1 is a diagram showing a configuration of an estimation server according to an embodiment of the present invention. In FIG. 1, only a necessary functional configuration is shown assuming a single computer system. Therefore, the estimation server may be a distributed server computer composed of a plurality of server computers.

  The estimation server 100 has a configuration in which a RAM 111, an input device 112, an output device 113, a communication control device 114, and a storage device 116 that is a nonvolatile storage medium (ROM, HDD, etc.) are connected to a CPU 110 via a system bus 115. Have The storage device 116 includes a program storage area that stores a software program for performing each function of the commitment line withdrawal rate estimation model, and a data storage area that stores data handled by the software program. Each means of the program storage area described below is actually an independent software program, its routines, components, and the like, which are called from the storage device 116 by the CPU 110 and expanded in the work area of the RAM 111, referring to the database and the like as appropriate. Each function is performed by being executed sequentially.

  Next, among the software programs stored in the program storage area in the storage device 116, only those related to the present invention are listed, the rating reduction range calculation means 120, the additional withdrawal rate calculation means 121, the data aggregation means 122, and A weighted average calculating means 123 is provided. These means are executed by the CPU 110.

  The rating decline calculation means 120 calculates a rating decline for each obligor. The rating downgrade is the number of stages in which a debtor's rating has been lowered, which is set according to the original regulations of a financial institution (for example, a rating of 13 levels, the highest rating being “1” and the lowest rating being “13”). (Default) ”, the rating drop when the rating is reduced from 1 to 13 is 12. When the estimation model is updated annually, the rate of decline between the rating one year ago (past) and the current rating is calculated.

  The additional withdrawal rate calculation means 121 extracts additional withdrawal performance data from a withdrawal performance data storage unit 130, which will be described later, and calculates the default withdrawal rate and the non-default rating decline destination commitment line additional withdrawal rate. Here, the default destination means a trading customer whose rating has become the default in the last one year. A non-default rating downgrader refers to a debtor whose rating has declined in the last year among debtors other than the default. The estimation model based on the default destination is the conventional one, and the estimation model using the non-default rating lowering destination is the estimation model according to the present invention.

  The data totaling means 122 averages the calculated commitment line additional withdrawal rate for each rating decrease width, and stores it in the total data storage unit 131 classified by rating decrease width, which will be described later. Further, the data totaling means 122 calculates PD values (expected values of default probabilities, see Table 1), the number of obligors, and the number of expected DFs for each rating, and stores them in a rating-based total data storage unit 132 described later. . Here, the expected number of DFs is the number of obligors whose ratings are expected to become default in the future (for example, one year later). The expected number of DFs can be calculated by PD value × number of obligors.

  The weighted average calculation means 123 calculates a weighted average value of the calculated commitment line additional withdrawal rate for each of the number of obligors by rating and the expected number of DFs by rating, and stores them in a weighted average data storage unit 133 to be described later. .

  Next, as for the data storage area in the storage device 116, only those related to the present invention are enumerated, the withdrawal result data storage unit 130, the total data storage unit 131 classified by rating down width, the total data storage unit 132 classified by rating, and the weighting An average data storage unit 133 is provided. Both are fixed storage areas secured in the storage device 116.

  The withdrawal result data storage unit 130 stores the result data of the withdrawal amount. FIG. 3 is a diagram showing data stored in the withdrawal result data storage unit according to the embodiment of the present invention. The withdrawal performance data in FIG. 3 is the performance data for each withdrawal made within the loan facility, and indicates the “obligor ID” that uniquely indicates the debtor, the “drawing date” that indicates the date of withdrawal, and the withdrawal amount. Stores “withdrawal amount”. Since this data is the actual data for each drawer, as shown in FIG. 3, there may be a plurality of identical IDs in the “obligor ID”.

  The total data storage unit 131 for each grade decline width stores average data of the commitment line additional withdrawal rate for each grade decline width. FIG. 4 is a diagram showing data stored in the aggregated data storage unit classified by rating reduction width according to the embodiment of the present invention. The aggregated data for each rating decline in Fig. 4 includes the "rating decline" indicating the number of stages in which the debtor's rating has declined, and the "drawing rate estimate" indicating the average value of the additional withdrawal rate for each commitment line. Is stored.

  The total data storage unit 132 according to rating stores data related to a customer who requests a transaction. FIG. 5 is a diagram showing data stored in the rating-based total data storage unit according to the embodiment of the present invention. The aggregated data by rating in FIG. 5 includes “rating” indicating the rating at the time of aggregation, “PD value” indicating the expected probability of bankruptcy shown in Table 1, and “number of obligors” indicating the number of obligors for each rating. , And “expected number of DFs” indicating the expected number of obligors whose ratings will become the default after a predetermined period (one year when the estimation model is updated annually). As described above, the “expected number of DFs” can be calculated by the number of obligors × PD value.

  The weighted average data storage unit 133 stores weighted average data of the commitment line additional withdrawal rate in each of the number of obligors by rating and the expected number of DFs. FIG. 6 is a diagram showing data stored in the weighted average data storage unit according to the embodiment of the present invention. The weighted average data in Figure 6 is the “weighted average number of obligors by rating” that shows the average of the commitment line additional withdrawal rate weighted by the number of obligors by rating, and the expected DF number by rating for the additional withdrawal rate by commitment line. The “expected DF number weighted average” indicating the average value weighted with is stored.

  Next, with reference to the flowcharts of FIGS. 7 and 8 and the data of FIGS. 3 to 6, the commitment line additional withdrawal rate aggregation processing according to the embodiment of the present invention will be described along the flow. FIG. 7 is a flowchart showing a commitment line additional withdrawal rate totaling process according to an embodiment of the present invention. This process is a batch process executed in the estimation server 100 for the purpose of establishing an estimation model of the commitment line additional withdrawal rate. This processing described below is an embodiment, and is assumed to be executed once a year in order to perform annual update of the estimation model.

  First, in step 701, the rating decline calculation means 120 acquires the rating of the debtor from debtor data (not shown). The ratings to be acquired are the current rating and the rating one year ago. Further, as shown in FIG. 5, the rating is a 13-level rating in which the highest rating is “1” and the lowest rating is “13 (default)”, which is also an embodiment. In this step, it is assumed that a rating for one obligor is obtained. That is, in this process, when the process for one obligor is completed, the process for the next obligor is executed, and it is assumed that the aggregation process is performed by repeating this process. However, in another embodiment, the ratings of all obligors can be acquired and the aggregation process can be performed.

  Next, the rating decline calculation means 120 calculates a rating decline from the difference between the current rating acquired in step 701 and the rating one year ago (step 702). For example, when the rating is reduced from 1 to DF, the rating decline is 12. If the current rating and the rating one year ago are the same, the rate of decline is 0, and if the rating is up, for example, the rate of decline is a negative value (for example, the rate of decline when going from 4B to 3 is- 2).

  Next, the rating reduction range calculation means 120 determines whether the debtor is a default destination, a non-default rating reduction destination, or others based on the rating reduction range calculated in step 702 (step 703). Here, the default destination is the debtor whose rating has become the default, although the rating has declined. A non-default rating downgrader is a debtor whose default rating is a rating other than the default, although the rating has declined. Others are obligors whose ratings have not changed from one year ago and those whose ratings have increased. The conventional estimation model targets only the default destination, and in this estimation model, the obligor whose rating has been lowered (that is, default destination + non-default rating down destination) is the estimation target, and thus such a determination is made.

  When the determination result in step 703 is the default destination or the non-default rating lowering destination (that is, when the rating is lowering), the process proceeds to the Yes route in step 704, and the additional withdrawal rate calculation means 121 extracts the withdrawal result data. Additional withdrawal result data within the last year of the debtor is extracted from the storage unit 130 (step 705). Specifically, for the withdrawal result data (FIG. 3), the additional withdrawal amount for the most recent year on a specific reference date (for example, the end of March) is acquired using the debtor ID of the subject debtor as a search key. If the performance data over multiple years can be acquired, the additional withdrawal amount that can be acquired will be multiple minutes. If there is no withdrawal history data within the last year, the additional withdrawal amount will be 0 yen.

  Next, the additional withdrawal rate calculation means 121 calculates a commitment line additional withdrawal rate based on the acquired additional withdrawal performance data (additional withdrawal amount) and the like (step 706). Specifically, as shown in Fig. 2, the commitment is obtained by dividing the "additional withdrawal amount", which is the current spent amount minus the used amount one year ago, by the "unused amount" one year ago. The line drawing rate can be calculated.

  After step 706, or when the determination result in step 703 is not the default destination or the non-default rating lowering destination (that is, when the rating has not changed or has been increased), the process proceeds to the Yes route of step 704 to calculate the rating lowering range. The means 120 determines whether or not there is a next debtor (step 707). If it is determined that there is a next debtor, the process proceeds to the Yes route, and the rating reduction range calculation means 120 acquires the rating of the next debtor from debtor data (not shown) (step 708). Note that the rating to be acquired is the same as that described in step 701. Thereafter, steps 702 to 708 are repeated until there is no next debtor.

  If it is determined in step 707 that the next obligor does not exist, the process proceeds to No route, and each data aggregation process is executed (step 709). The data totaling process in step 709 will be described later with reference to FIG. After step 709, the process ends.

  Next, details of the data totaling process in step 709 will be described with reference to the flowchart of FIG. FIG. 8 is a flowchart showing details of the commitment line additional withdrawal rate tabulation process according to the embodiment of the present invention. This processing is targeted for debtors determined to be default destinations, non-default rating destinations in step 704 (that is, debtors whose ratings have declined).

First, in step 801, the data totaling means 122 averages the commitment line additional withdrawal rate calculated in step 706 for each rating decrease range calculated in step 702, and creates aggregated data by rating decrease range (FIG. 4). Then, the data is stored in the aggregate data storage unit 131 according to the rating decline width. The “drawing rate estimate value” in FIG. 4 is the average value of the commitment line additional withdrawal rate for each rating decline, and this can be used as one index for each rating decline in this estimation model. It should be noted that there is a possibility that the number of additional withdrawals of commitment lines (that is, the number of obligors) used at step 801 for each rating decline is not sufficient. Therefore, it is also possible to formulate the relationship between each rating decline and the commitment line additional withdrawal rate corresponding to each rating decline, and correct (estimate) the withdrawal rate estimate by regression analysis. For example, by regression analysis, a relational expression such as exponential function y = ae ^bx (x: commitment line additional withdrawal rate, y: rating) from each grade decline and the average value of the additional commitment line withdrawal corresponding to each grade decline (Decrease width). By applying each rating decline to this relationship, it is possible to estimate the average value of the commitment line additional withdrawal rate corrected for each rating decline. This estimated value can be used in a future step as a withdrawal rate estimated value.

  Next, the data aggregation means 122 calculates the PD value (expected value of probability of bankruptcy, see Table 1), the number of obligors, and the expected number of DF for each rating, and creates aggregated data by rating (FIG. 5). Then, it is stored in the total data storage unit 132 classified by rating (step 802). The “PD value” in FIG. 5 is a value estimated for each rating by the financial institution. The “number of obligors” in FIG. 5 is the total number of obligors for each rating. The “expected DF value” in FIG. 5 can be calculated by PD value for each rating × number of obligors. Note that the processing order of steps 801 and 802 may be reversed.

  Next, the weighted average calculating means 123 performs weighted averaging of the withdrawal rate estimated value calculated in step 801 by each of the number of obligors by rating and the expected number of DFs by rating to create weighted average data (FIG. 6). And stored in the weighted average data storage unit 133 (step 803). The "weighted average number of obligors by rating" in Fig. 6 is calculated by calculating the withdrawal rate estimate (Fig. 4) x the number of obligors (Fig. 5) for each rating. The total of each calculated value (75% x 100 companies + 60%) × 500 companies + 50% × 1000 companies + 40% × 2000 companies + 35% × 3000 companies +... + 5% × 3000 companies) can be calculated by dividing by the total number of debtors. “Expected DF value weighted average” is calculated as the withdrawal rate estimated value (Fig. 4) x expected DF number (Fig. 5) for each rating, and the total of each calculated value (75% x 0 companies + 60% x 0 companies) + 50% × 5 companies + 40% × 10 companies + 35% × 10 companies +... + 5% × 1000 companies) is divided by the total number of expected DFs. After step 803, the process ends and returns to the flowchart of FIG.

  The weighted average number of obligors by rating (20.36%) and the expected DF value weighted average (9.74%) of the commitment line additional withdrawal rate calculated in Step 803 are the estimated values in this estimation model. By using these estimated values as input parameters (commitment line additional withdrawal rate) when calculating the capital adequacy ratio, it is possible to calculate highly stable estimated values.

  As described above, according to the present invention, according to the present invention, it is possible to calculate a more stable commitment line additional withdrawal rate (as a result, capital adequacy ratio) as an estimated value, enabling classification estimation by segment, and the present invention. Can be used as a benchmark model to confirm the validity of the current estimation method.

Claims (6)

A computer for estimating an additional withdrawal rate of a commitment line in order to calculate a capital ratio of a financial institution,
Obtain ratings of multiple obligors, the ratings are set by the financial institution in accordance with its own provisions, and include past and current ratings,
From the difference between the previous rating and the current rating, calculate a rating decline in a predetermined period from the past to the present,
On the basis of the rating reduction range, among the plurality of debtors, determine a debtor whose rating has declined in the predetermined period,
For the debtor whose rating has declined, calculate the additional withdrawal amount from the credit line in the predetermined period,
Based on the additional withdrawal amount, a commitment line additional withdrawal rate is calculated,
The computer is configured to calculate an average value of the additional withdrawal rate of the commitment line for each of the rating reduction widths.   Regression analysis formulates the relationship between the rating decline width and the average value of the commitment line additional withdrawal rate, and the commitment line is added by inputting the rating decline width into the formulated relational expression. The computer according to claim 1, further configured to correct the average value of the withdrawal rate and to estimate a withdrawal rate estimated value.   The withdrawal rate estimate is further configured to be weighted average by the number of obligors by rating or the expected number of DFs, and the expected number of DFs is the number of obligors that are expected to default in the future. The computer according to claim 2.   The computer according to claim 3, wherein the expected number of DFs is calculated by multiplying the number of obligors by rating and a PD value. A computer-implemented method for estimating a commitment line additional withdrawal rate to calculate a capital institution's capital ratio, the method comprising:
Obtaining a rating of a plurality of debtors, wherein the rating is set by the financial institution in accordance with its own provisions and includes a past rating and a current rating;
A step of calculating a rating decline in a predetermined period from the past to the present from a difference between the past rating and the current rating;
Determining a debtor whose rating has declined during the predetermined period among the plurality of debtors based on the rating decline width;
Calculating an additional withdrawal amount from the credit line in the predetermined period for the debtor whose rating has declined;
Calculating a commitment line additional withdrawal rate based on the additional withdrawal amount;
And calculating an average value of the additional withdrawal rate of the commitment line for each of the rating decline widths. A computer program that causes a computer to execute a computer-implemented method of estimating a commitment line additional withdrawal rate to calculate a capital institution ratio of a financial institution, and the computer program is executed by the computer, In the computer,
Obtaining a rating of a plurality of debtors, wherein the rating is set by the financial institution according to its own provisions, and includes a past rating and a current rating;
From the difference between the previous rating and the current rating, let us calculate the rating decline in the predetermined period from the past to the present,
Based on the rating reduction range, among the plurality of debtors, the debtor whose rating has declined in the predetermined period is determined,
Let the debtor whose rating declines calculate the additional withdrawal amount from the credit line in the predetermined period,
Based on the additional withdrawal amount, the commitment line additional withdrawal rate is calculated,
An average value of the commitment line additional withdrawal rate is calculated for each rating reduction range.

Images