JP2018206023A - Risk evaluation system, risk evaluation method and risk evaluation program - Google Patents

Abstract

To provide a risk evaluation system, a risk evaluation method, and a risk evaluation program for efficiently evaluating an interest rate risk in arbitrary institution.SOLUTION: A control unit 21 in a support server 20 generates, in order to calculate interest rates and institutional indicators using released nationwide information, an interest rate prediction model and an index prediction model including a chain graphical model having a hierarchical Bayesian structure; adjusts, when individual summary information for calculating an institutional index is acquired from a specific institution, the index prediction model using individual summary information; calculates, according to an interest rate calculated based on the interest rate prediction model, a predicted value of the institutional index using the adjusted index prediction model; calculates a risk index based on the predicted value of the institutional index; and performs Bayesian update for the index prediction model using newly released nationwide information or newly acquired individual summary information.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a risk evaluation system, a risk evaluation method, and a risk evaluation program for evaluating interest rate risk.

  Interest rate fluctuations affect bond prices. Therefore, a technique for evaluating interest rate risk for bonds (bonds) whose cash flow changes according to interest rate fluctuations has been studied (see, for example, Patent Document 1). The interest rate risk evaluation program described in this document calculates the cash flow of a bond using a mathematical model that represents a time-varying path of interest, and calculates the effective duration of the bond using the cash flow.

  Also, Bayesian estimation may be used to predict the future. In this Bayesian estimation, based on the concept of Bayesian probabilities, the matter to be estimated (cause event that is the cause) is inferred in a probabilistic sense from the observed event (observed fact). A calculation method of a bankruptcy probability that can calculate a realistic probability irrespective of the ratio of bankruptcy / non-bankruptcy in the learning data has also been studied (for example, see Patent Document 2). In the technology described in this document, the financial data of multiple companies are input, a decision tree is created from the financial data, the financial data of the target company is applied to the decision tree, Bayes' theorem is used, and the target company is The probability of default is calculated from the result of applying to the decision tree. Furthermore, a hierarchical Bayes model in which the relationship between parameters is described using a hierarchy of prior distribution and hyper prior distribution has been studied.

Japanese Patent Laid-Open No. 2006-162417 JP 2000-259719 A

  In April 2016, the Basel Committee on Banking Supervision published “Standards Interest rate risk in the banking book (IRRBB)”. Each financial institution is scheduled to respond to this IRRBB. IRRBB reviews the bank supervisory treatment of interest rate risk for bank accounts (principle for interest rate risk management and supervision) and requires stricter interest rate risk management.

For each financial institution that needs to respond, for the purpose of interest rate risk management, estimates are made with fixed-rate loans with prepayment risk, fixed-rate loan commitments, time deposits with early termination risk, and deposits with no maturity (NMDs). The model needs to be examined.
Bank operations are classified into bank accounts that manage transactions centered on loans and deposits, and trading accounts that manage transactions for the purpose of securing short-term profit margins due to fluctuations in interest rates. Bank accounts include loans, deposits, and government bonds intended to be held to maturity, and trading accounts include receivables for the purpose of earning profits from daily trading. For interest rate risk in the trading book, it was necessary to accumulate certain capital. Furthermore, it became necessary to strengthen the supervisory function for government bonds held in bank accounts.
Banking interest rate risk is the risk arising from fluctuations in the market value or revenue of banking account assets and liabilities due to changes in interest rate levels. Bank accounts are basically operated for the long-term holding of assets, etc., and in terms of regulations, risk of credit loss (= credit risk) rather than changes in market prices due to changes in interest rates (= interest rate risk) Has been emphasized.

However, the evaluation of the elements related to behavioral options required a large amount of data considering various situations, and the evaluation was difficult.
In addition, in order to create a statistical model that can be used by a plurality of arbitrary financial institutions, it is convenient to have a mechanism that allows rational estimation even if the data structure is unknown. In addition, it is not realistic to obtain contract information and the like from financial institutions in advance and create a statistical model dedicated to each financial institution each time. That is, it is difficult to ensure the estimation accuracy when inputting unknown data with a statistical model constructed with static data of a specific financial institution.

  A risk evaluation system that solves the above problems includes a public information storage unit that stores public nationwide information, an individual information storage unit that stores individual summary information that represents an index in a specific organization, and a control unit that calculates a risk index With. And the said control part produces | generates the interest rate prediction model and index prediction model which consist of the chain | strand-shaped graphical model which has a hierarchical Bayesian structure, in order to calculate an interest rate and an organization index using the publicly disclosed national information, and a specific organization When the individual summary information for calculating the engine index is acquired from the above, the index prediction model is adjusted using the individual summary information, and the adjusted index is adjusted according to the interest rate calculated based on the interest rate prediction model. An engine index prediction value is calculated using the prediction model, and Bayesian update is performed on the index prediction model using newly disclosed national information or newly acquired individual summary information.

  The present invention has been made to solve the above-described problems, and can effectively evaluate the interest rate risk in an arbitrary organization.

Explanatory drawing of the risk evaluation system of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the information used by this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment. Explanatory drawing of the process sequence of this embodiment.

  Hereinafter, an embodiment of a risk evaluation system embodying the present invention will be described with reference to FIGS. In the present embodiment, it is assumed that IRRBB is evaluated using a hierarchical Bayes model. Here, a chain-like graphical model (for example, a Bayesian network) is used as the hierarchical Bayes model (Bayes statistical model). This chain-like graphical model consists of two types of models called “system models” that express time-series fluctuations in data and “observation models” that calculate target information (latent variables) from data at a certain point in time. Composed. In the present embodiment, this model is constructed using Bayesian statistics.

As shown in FIG. 1, in this embodiment, a client terminal 10 and a support server 20 are used.
The client terminal 10 is a computer terminal used by a user who uses this service. The client terminal 10 includes a control unit, an output unit, and an input unit. The control unit executes various applications. In the present embodiment, the control unit functions as a browser for accessing the support server 20. The output unit includes a display and outputs various information. The input unit is configured by a keyboard and a pointing device, and acquires an instruction input by the user.

The support server 20 is a computer system that evaluates interest rate risk using a hierarchical Bayesian model. The support server 20 includes a control unit 21, a public information storage unit 22, and an individual information storage unit 23.
The control unit 21 includes control means (CPU, RAM, ROM, etc.) and performs processing (interest rate prediction stage, core deposit prediction stage, fixed deposit prediction stage, advanced repayment rate prediction stage, risk calculation stage, hedge simulation stage, which will be described later. Each process). By executing the risk evaluation program for this purpose, the control unit 21 causes the interest rate prediction unit 211, the core deposit prediction unit 212, the term deposit prediction unit 213, the advanced repayment rate prediction unit 214, the risk calculation unit 215, and the hedge simulation unit 216. Function as.
The interest rate prediction unit 211 performs future prediction of a zero coupon / yield curve, and executes a process of calculating a discount rate.
The core deposit prediction unit 212 executes a process of calculating the core deposit balance in the shock scenario using a hierarchical Bayes model.
The time deposit prediction unit 213 executes a process of calculating a time deposit amount (balance) in the shock scenario using a hierarchical Bayes model.
The advance repayment rate prediction unit 214 executes a process of calculating the advance repayment rate in the shock scenario using a hierarchical Bayesian model.

The hierarchical Bayesian model used by the core deposit prediction unit 212 to the advanced repayment rate prediction unit 214 is generated using individual information and public information. When new individual information and public information are acquired, the hierarchical Bayes model is updated.
The risk calculation unit 215 executes processing for calculating interest rate risk using the estimated interest rate, core deposit, fixed deposit amount, and advanced repayment rate.
The hedge simulation unit 216 executes a hedge simulation process corresponding to the interest rate risk.

The public information storage unit 22 records comprehensive public information (here, national information). Public information is recorded when public information is acquired. In the public information, information necessary for generating a hierarchical Bayes model for calculating interest rates, core deposits, fixed deposit amounts, and advanced repayment rates is recorded.
Individual information in the financial institution is recorded in the individual information storage unit 23. Individual information is recorded when individual information is acquired from a financial institution. In this individual information, information (individual summary information) for adjusting the hierarchical Bayesian model generated using the public information is recorded.

(Outline of processing procedure)
The outline of the processing procedure will be described with reference to FIG. Here, a hierarchical Bayes model is created with comprehensive information, and IRRBB is predicted based on partial information.
First, the control unit 21 of the support server 20 executes a model creation process assuming every scene (step S1-1). Here, publicly disclosed overall information (national information) is used.

Next, the control unit 21 of the support server 20 executes model adjustment processing using the individual information (step S1-2). Here, individual information of a specific financial institution performing risk assessment is used.
Next, the control part 21 of the support server 20 performs the prediction process by the adjusted model (step S1-3). Here, the interest rate, core deposit, fixed deposit amount, and advanced repayment rate are calculated using a hierarchical Bayesian model.
Next, the control part 21 of the support server 20 performs the acquisition process of the performance corresponding to prediction (step S1-4).
Next, the control part 21 of the support server 20 performs a model update process (step S1-5). Here, when the disclosed comprehensive information (update information) is acquired, the hierarchical Bayes model is updated.

(Outline of data used in this system)
The outline of modules and data used in this system will be described with reference to FIG.
First, the interest rate prediction unit 211 uses the zero coupon / yield curve creation logic 31 to calculate a zero coupon / yield curve and a discount factor using public information. Here, the zero coupon yield curve and discount factor of each currency specified in the IRRBB regulation are calculated. Then, the interest rate prediction unit 211 converts the zero-coupon yield curve of the Japanese yen calculated from the zero-coupon yield curve and the discount factor and the macroeconomic index as public information into the interest-rate prediction model 32 ( Hierarchical Bayes model). Thereby, the probability distribution of the zero coupon yield at an arbitrary time is calculated.

  Next, the core deposit prediction unit 212 calculates the probability distribution of zero coupon yield at an arbitrary time, the historical liquid deposit interest rate, and the individual / corporate liquid deposit balance into the core deposit estimation model 33 (hierarchical Bayes model). To introduce. Thereby, the core deposit amount for each period bucket is calculated.

  Next, the time deposit forecasting unit 213 calculates the probability distribution of zero coupon yield at an arbitrary time point, the core deposit amount for each period bucket, the time deposit interest rate by historical contract term, and the balance from the time deposit amount estimation model 34 ( Hierarchical Bayes model). Thereby, the fixed deposit amount for every period bucket is calculated.

Furthermore, the advance repayment rate prediction unit 214 introduces the total principal, the number of contracts, the average value, the standard deviation, and the quantile for each period bucket into the fixed interest rate loan advance repayment rate estimation model 35 (hierarchical Bayes model). Thereby, the advance repayment rate for every period bucket is calculated.
The risk calculation unit 215 introduces the core deposit amount for each period bucket, the fixed deposit amount for each period bucket, and the prepayment rate for each period bucket into the interest rate risk index calculation module 36. Thereby, a risk evaluation result is calculated. As the risk assessment results, the overall portfolio interest rate, net asset fluctuation (ΔEVE) by shock scenario, and net interest income fluctuation (ΔNII) are calculated.

  Then, the hedge simulation unit 216 uses the hedge position simulator 37 to execute a simulation process.

(Zero Coupon Yield Curve Creation Logic)
Next, the zero coupon yield curve creation logic will be described with reference to FIG. Here, the future prediction of the zero coupon yield curve of the Japanese yen currency is performed, and this yield curve is developed into any currency.

  First, the control unit 21 of the support server 20 executes processing for acquiring information on a swap rate curve and a currency swap rate (step S2-1). Specifically, the interest rate prediction unit 211 of the control unit 21 acquires information on a swap rate curve and a currency swap rate prepared in advance.

  Next, the control part 21 of the support server 20 performs the creation process of a zero coupon coupon yield curve for every currency (step S2-2). Specifically, the interest rate prediction unit 211 of the control unit 21 creates a zero coupon / yield curve for each currency to be considered in the calculation of IRRBB from the acquired information. Next, the interest rate prediction unit 211 calculates a discount rate for each period bucket.

  Next, the control unit 21 of the support server 20 executes a discount rate calculation process for each period bucket (step S2-3). Specifically, the interest rate prediction unit 211 of the control unit 21 repeats the calculation of the discount rate from the zero-coupon yield curve of each currency using an interest rate prediction model described later. Note that if the zero coupon yield curve is estimated based on the OIS (next-day interest rate swap) swap rate maturity, the B-spline approximation is used to calculate the value of the central maturity of the period bucket. Perform linear interpolation with.

(Interest rate prediction model)
Next, the interest rate prediction model will be described with reference to FIG. In this interest rate prediction model, the future prediction of the zero coupon yield curve of the Japanese yen currency at a certain point in time is performed, and the yield curve is developed into an arbitrary currency.

  First, the control unit 21 of the support server 20 executes a macroeconomic index capturing process (step S3-1). Specifically, the interest rate prediction unit 211 of the control unit 21 acquires a macroeconomic index used for the hierarchical Bayes model from the public information storage unit 22.

  Next, the control unit 21 of the support server 20 executes a process of taking in the zero-coupon / yield curve of the Japanese yen at the start of prediction (step S3-2). Specifically, the interest rate prediction unit 211 of the control unit 21 obtains the calculated Japanese yen zero coupon yield curve in step S2-2.

Next, the control part 21 of the support server 20 performs the prediction process of the future zero coupon yield curve (step S3-3). Specifically, the interest rate prediction unit 211 of the control unit 21 predicts a future interest rate curve from the hierarchical Bayesian model. Here, a hierarchical Bayes model using the Nelson-Siegel model is used. The parameters are β_1, β_2, β_3 for expressing the level, slope, and curvature, respectively, λ required as common parameters, elapsed months m (m does not need to be estimated), β_1, β_2, β_3 are the following linear Gaussians Estimate by model. “_” Means a subscript.
State variable x_t, where H is the value obtained from λ calculated by the Nelson-Siegel model and the number of months elapsed, macroeconomic indicators (base loan rate, consumer price index (excluding fresh products) compared to the same month last year, US financial securities (10 years) Yield, US FF rate, TSE Stock Price Index, Industrial Production Index compared to the same month last year). If the abbreviations are BDR_t-1, CPI_t-1, US_10t-1, FF_t-1, TOPIX_ (t-1, IIP_ (t-1)), β_1t, β_2t, and β_3t can be described by the following equations.

  At this time, the hyperparameters are error terms ω_t, υ_t, μ_t, α, γ included in the state model, and a to f that act as “coefficients for explanatory variables” of the state model. ω_t, υ_t, and μ_t are each assumed to be a normal distribution N (0, 10) assuming no information. The other hyperparameters (assuming hp) when the data observation start is set to t = 1 are calculated by the following method. This reflects the time series structure in which the probability distribution of the parameter in the previous period is applied as the average of the probability distribution of the parameter in the period to be estimated, except for the observation start period. “^” Means a power.

  The parameter λ follows a normal distribution. The estimation is performed with the setting reflecting the same time series structure as the hyper parameters other than the error term in the estimation of the parameter β.

  Next, the control part 21 of the support server 20 performs the expansion | deployment process to arbitrary currencies (step S3-4). Specifically, the interest rate prediction unit 211 of the control unit 21 calculates a value for each period bucket in an arbitrary currency using a zero coupon / yield curve creation logic.

  Next, the control unit 21 of the support server 20 executes a discount rate calculation process (step S3-5). Specifically, the interest rate prediction unit 211 of the control unit 21 calculates a discount rate for an arbitrary currency.

(Core deposit estimation model)
Next, the core deposit estimation model will be described with reference to FIG. In this core deposit estimation model, the amount of deposits (core deposits) that remain without being withdrawn from liquid deposits changes according to changes in market interest rates. We estimate this by dividing it into individual deposits (further classified into non-tradable and transactable) and corporate deposits.

  First, the control unit 21 of the support server 20 executes historical liquid deposit interest rate acquisition processing (step S4-1). Specifically, the core deposit prediction unit 212 of the control unit 21 outputs a message prompting the client terminal 10 to upload a historical (assuming about five years) liquid deposit interest rate. Then, the core deposit prediction unit 212 records the historical liquid deposit interest rate uploaded from the client terminal 10 in the individual information storage unit 23.

  Next, the control unit 21 of the support server 20 executes an individual / corporate-specific liquid deposit balance acquisition process (step S4-2). Specifically, the core deposit prediction unit 212 of the control unit 21 outputs a message prompting the client terminal 10 to upload the individual / corporate liquid deposit balance. Then, the core deposit prediction unit 212 records the individual / corporate-specific liquid deposit balance uploaded from the client terminal 10 in the individual information storage unit 23.

Next, the control part 21 of the support server 20 performs the process of estimating the market following rate of the liquid deposit interest rate (step S4-3). Specifically, the core deposit prediction unit 212 of the control unit 21 uses the historical data recorded in the individual information storage unit 23 and a market interest rate tracking rate model (hierarchical base model described later) of the liquid deposit interest rate. Estimate the market following rate of sex deposit interest rate.
Here, we use the monthly zero coupon rate as the market funding rate, and use the market funding rate, the liquid deposit rate (such as the ordinary deposit rate) and the tracking rate of the liquid deposit rate and the liquid deposit rate and the fixed deposit rate. Assume a replicating portfolio that matches the interest rate of interest. By solving this, a hierarchical Bayesian model for estimating the following rate θ is generated and expressed by the following equation.

  The tracking rate θ is a model reflecting the following time-series structure with a normal distribution with mean μ and standard deviation as hyperparameters, and both hyperparameters with t = 1 as the data start period (summarizing hyperparameters). Ph).

  In addition, the liquidity interest rate in the future can be calculated by changing the market procurement interest rate r_m ^ t using this hierarchical Bayesian model.

  Next, the control part 21 of the support server 20 performs the prediction process of a Japanese yen zero coupon coupon yield curve (step S4-4). Specifically, the core deposit prediction unit 212 of the control unit 21 estimates the posterior distribution of the probability distribution of the liquid deposit interest rate from the calculated market follow rate of the liquid deposit interest rate and the interest rate prediction model.

Next, the control unit 21 of the support server 20 executes a posterior distribution estimation process for the liquid deposit balance (step S4-5). Specifically, the core deposit prediction unit 212 of the control unit 21 uses the calculated Japanese yen zero coupon coupon yield curve and a model for predicting the future level of liquid deposits (hierarchical base model to be described later). Estimate the posterior distribution parameters of the balance.
Here, the liquid deposit amount M_t in the future period t is expressed by a hierarchical Bayesian model based on the AR1 structure.

  The parameter κ is affected by the liquidity rate, macroeconomic indicators, and adjustment terms set for each financial institution. If the macroeconomic index is ρ_i, it can be expressed by the following formula.

  Hyperparameters a and b_i, which function like the constant term I and the macroeconomic index coefficient, are as follows when the data observation start is set to t = 1 (assuming that it is hp. Calculate by the method. This reflects the time series structure in which the probability distribution of the parameter in the previous period is applied as the average of the probability distribution of the parameter in the period to be estimated, except for the observation start period.

  Next, the control part 21 of the support server 20 performs the estimation process of the core deposit amount by a model (step S4-6). Specifically, the core deposit prediction unit 212 of the control unit 21 sets the 99% lower limit of each period bucket calculated from the posterior distribution as the core deposit amount for the period.

  Next, the control part 21 of the support server 20 performs the adjustment process of a period bucket distribution rate (step S4-7). Specifically, the core deposit prediction unit 212 of the control unit 21 calculates a core deposit distribution rate for each period bucket from the calculated core deposit amount for each period bucket, and calculates an average remaining period of the core deposits therefrom. When this value exceeds the average remaining period upper limit of the deposit type stipulated in the regulations, the core deposit distribution ratio of the longest period bucket is decreased in order, and the average remaining period is adjusted so as to match the regulatory upper limit.

  Next, the control part 21 of the support server 20 performs the adjustment process of the total amount of core deposits (step S4-8). Specifically, the core deposit prediction unit 212 of the control unit 21 determines that the total amount of core deposits obtained by adding up the estimated core deposit amounts by period bucket exceeds the upper limit of the core deposit amount ratio in the liquid deposits defined in the regulations. The core deposit amount is decreased so that the core deposit amount ratio matches the upper limit.

  Next, the control part 21 of the support server 20 performs the distribution process of the core deposit to a period bucket (step S4-9). Specifically, the core deposit prediction unit 212 of the control unit 21 multiplies the core deposit total adjusted in step S4-8 by the core deposit distribution rate by period bucket adjusted in step S4-7, and sets each period bucket. To distribute core deposits.

(Time deposit amount estimation model)
Next, the time deposit amount estimation model will be described with reference to FIG. In this time deposit amount estimation model, the balance of time deposits fluctuates according to changes in market interest rates. This is estimated by contract term, and the cancellation amount of time deposits is calculated for each period bucket and redistributed to the appropriate period bucket.

  First, the control unit 21 of the support server 20 executes processing for acquiring historical fixed deposit interest rates by contract term and the same balance (step S5-1). Specifically, the time deposit prediction unit 213 of the control unit 21 outputs a message prompting the client terminal 10 to upload the term deposit interest rate by contract term and the term deposit balance by contract term. Then, the time deposit prediction unit 213 records in the individual information storage unit 23 the historical time deposit interest rates by contract term and the term deposit balance by contract term uploaded from the client terminal 10.

  Next, the control part 21 of the support server 20 performs the calculation process of a time deposit interest rate following rate (step S5-2). Specifically, the time deposit prediction unit 213 of the control unit 21 uses the historical data uploaded by the user and the market interest rate follow-up rate model of the time deposit interest rate (hierarchical Bayes model described later), the market follow-up rate of the time deposit interest rate. Is estimated.

  Procurement consisting of fixed deposit interest rate and fixed interest rate using market coupon interest rate, fixed deposit interest rate and follow-up rate of fixed deposit interest rate with zero coupon rate of period bucket including term of contract of fixed deposit pool as market procurement interest rate Assume a replicating portfolio that matches interest rates. This is a hierarchical Bayesian model that estimates the following rate θ by solving this, and is expressed by the following equation.

  The tracking rate θ is a model reflecting the following time-series structure, in which the following parameters are represented by a normal distribution with mean μ and standard deviation as hyperparameters, and both hyperparameters have a data start period of t = 1. ph).

  In addition, using this hierarchical Bayesian model, the future fixed deposit interest rate can be calculated by changing the market procurement interest rate r_m ^ t.

  Next, the control part 21 of the support server 20 performs the prediction process of a Japanese yen zero coupon coupon yield curve (step S5-3). Specifically, the time deposit prediction unit 213 of the control unit 21 estimates the posterior distribution of the probability distribution of the time deposit interest rate from the calculated time deposit interest rate following rate and the interest rate prediction model.

Next, the control unit 21 of the support server 20 executes a posterior distribution estimation process for the time deposit balance (step S5-4). Specifically, the time deposit prediction unit 213 of the control unit 21 estimates the posterior distribution parameter of the time deposit balance using the predicted Japanese yen zero coupon coupon yield curve and the time deposit amount model (hierarchical Bayes model to be described later). To do.
The fixed deposit amount M_t in the future period t is expressed by a hierarchical Bayesian model based on the AR1 structure.

  The parameter κ is affected by the fixed deposit interest rate, the macroeconomic index, and the adjustment term set for each financial institution. If the macroeconomic index is ρ_i, it can be expressed by the following formula.

  Hyperparameters a and b_i, which function like the constant term I and the macroeconomic index coefficient, are as follows when the data observation start is set to t = 1 (assuming that it is hp. Calculate by the method. This reflects the time series structure in which the probability distribution of the parameter in the previous period is applied as the average of the probability distribution of the parameter in the period to be estimated, except for the observation start period.

  Next, the control unit 21 of the support server 20 executes a process for distributing the fixed deposit amount to the period bucket (step S5-5). Specifically, the fixed deposit prediction unit 213 of the control unit 21 calculates the fixed deposit amount of the period bucket by counting the fixed deposit amount by elapsed period for each period bucket from the posterior distribution.

(Fixed rate loan prepayment rate estimation model)
Next, a fixed interest rate loan advance repayment rate estimation model will be described with reference to FIG. In this fixed interest rate loan advance repayment rate estimation model, the advance repayment rate of positions where the incentive for advance repayment is likely to change due to changes in market interest rates, among fixed rate loans, is estimated. The specific positions assumed include relatively long-term certificate loans such as mortgages and investment mortgages.

  Note that business loans that simply cannot be refinanced are not included from the standpoints of extreme frame transactions such as card loans and maintaining relationships with financial institutions.

  First, the control unit 21 of the support server 20 executes acquisition processing of summary data of the total principal, the number of contracts, the average value, the standard deviation, and the quantile for each data item (step S6-1). Specifically, the advance repayment rate prediction unit 214 of the control unit 21 outputs a message prompting the client terminal 10 to upload summary data. The summary data includes the following data items for each period bucket corresponding to the contractual remaining period (maturity of contract repayment for fixed-term fixed-term contracts, interest rate change period for fixed-period fixed-rate or variable interest rates). Is used. Here, for each data item, the total principal, number of contracts, average value, standard deviation, quantile (10% point, 20% point, 30% point, 40% point, 50% point, 60% point, 70% Summary data of% point, 80% point, and 90% point) are used. Data items include principal balance, contracted interest rate, remaining months, annual income at the time of application, self-funding ratio at the time of application, repayment burden ratio at the time of application, and the prefecture where the property is located. Then, the advance repayment rate prediction unit 214 records the summary data uploaded from the client terminal 10 in the individual information storage unit 23.

Next, the control unit 21 of the support server 20 executes a parameter estimation process for the advanced repayment rate estimation model (step S6-2). Specifically, the advance repayment rate prediction unit 214 of the control unit 21 updates the probability distribution of parameters estimated in advance from the observation data of the Housing Finance Support Organization with the data uploaded by the user.
Here, it is known in practice that the time series transition of the advanced repayment rate is a process close to a lognormal distribution. Therefore, a hierarchical Bayesian model assuming a lognormal distribution is constructed.
The advanced repayment rate (CPR) in the elapsed years t is a function composed of a probability distribution (dCPR) and the elapsed years t.

  The average (expected value) of this dCPR is m, and the variance is v. At this time, the relationship between the log-normal distribution parameters μ and θ, the expected value m, and the variance v can be expressed as follows.

  Expected value m is a linear combination model that uses constant term I and macroeconomic index, annual income at the time of application, ratio of self-funding at the time of application, repayment burden ratio at the time of application, property location, etc. (w_1 to w_j in the formula). is there.

  Furthermore, it is assumed that there is an adjustment term (υ_i) due to the difference between financial institutions, and the variance v follows a normal distribution. Therefore, I, c, υ_i and μ ′, θ ′ are all hyper parameters. Among these, υ_i sets a normal distribution without information, and μ ′ and θ ′ set a gamma distribution without information. The other hyperparameters when the data observation start is set to t = 1 (assuming that it is hp. Read for each hyperparameter) are calculated by the following method. This reflects the time series structure in which the probability distribution of the parameter in the previous period is applied as the average of the probability distribution of the parameter in the period to be estimated, except for the observation start period.

  Next, the control unit 21 of the support server 20 uses the updated advanced repayment estimation model to execute an estimation process of the advanced repayment rate for each period bucket (step S6-3). Specifically, the advanced repayment rate prediction unit 214 of the control unit 21 estimates the probability distribution of the advanced repayment rate for each period bucket, and calculates the average value (CPR ′). Then, using the maturity y of the period bucket for which the advance repayment rate is to be calculated, the result of the following calculation is defined as the advance repayment rate (CPR).

  Next, the control part 21 of the support server 20 performs the correction process of the principal by period bucket (step S6-4). Specifically, the advance repayment rate prediction unit 214 of the control unit 21 subtracts the amount obtained by multiplying the advance repayment rate of the previous period by the principal from the principal for each period calculated based on the contract repayment. Calculated as principal after taking into account prepayments. Here, all buckets are calculated in order.

(Interest rate risk index calculation module)
Next, the interest rate risk index calculation module will be described with reference to FIG. The interest rate risk index calculation module implements a standardized platform-based approach described in the final Basel Committee document. According to the user's selection, it is possible to select a position calculation method having action options.

  First, the control unit 21 of the support server 20 executes a totaling position sorting process (step S7-1). Specifically, the risk calculation unit 215 of the control unit 21 classifies all positions in the banking account that are affected by the interest rate risk into three categories: standardization, nonstandardization, and nonstandardization.

  Next, the control unit 21 of the support server 20 executes position classification processing with interest rate option (step S7-2). Specifically, the risk calculation unit 215 of the control unit 21 classifies positions that have automatic interest rate option characteristics that are clearly defined in a contract or the like out of standardization failure. However, this position is subject to various calculations as a position with no interest rate option even in the following calculations.

  Next, the control part 21 of the support server 20 performs the division | segmentation process to a period bucket (step S7-3). Specifically, the risk calculation unit 215 of the control unit 21 classifies interest rate risk measurement target positions into period packets. Here, a period packet is set based on the principal repayment (contractual maturity), the interest rate renewal period, and the interest payment period for other tranches.

  Next, the control unit 21 of the support server 20 performs a zero coupon yield curve and discount rate calculation process (step S7-4). Specifically, the risk calculation unit 215 of the control unit 21 uses a zero coupon / yield curve creation logic.

  Next, the control unit 21 of the support server 20 executes a cash flow calculation process of an action option position (step S7-5). Specifically, the risk calculation unit 215 of the control unit 21 calculates cash flows for the core deposit, the fixed deposit amount, the advanced repayment rate, and the pull / throw rate.

<Core deposit>
Regarding core deposits, first, the risk calculation unit 215 classifies deposits with no maturity (NMD) as follows.
・ Retail deposit account (settlement): Regular deposit account or other non-profit account in retail deposits ・ Retail deposit (non-settlement): Retail deposit account that is not trading account ・ Wholesale Retail deposit account If it is difficult to discriminate between settlement and non-settlement, it is treated as a non-settlement account with a conservative upper limit on the core deposit ratio.
When the internal model is not used, the estimated value input by the user is used. When the internal model is used, NMD is allocated to each period bucket according to the core deposit amount calculated by the core deposit estimation model described above. To do.

<Time deposit amount>
When the internal model is not used, the risk calculation unit 215 calculates a TDRR parameter using a scenario multiplier for each interest rate shock scenario and a midterm lapse rate (TDRR) estimated by a financial institution. The risk calculation unit 215 calculates cash flow by multiplying the time deposit balance by TDRR. On the other hand, when using the internal model, the risk calculation unit 215 estimates the cash flow using the above-described time deposit amount estimation model.

<Progression repayment rate>
When estimating the cash flow of a fixed interest rate loan with prepayment risk without using an internal model, the risk calculator 215 uses the scenario multiplier for each interest rate shock scenario and the actual advance repayment rate provided by the user. Calculate the advance repayment rate by bucket.
When estimating using the internal model, the risk calculation unit 215 uses the prepayment rate estimated using the above-described fixed interest rate loan prepayment rate estimation model.
Then, the risk calculation unit 215 adds the value obtained by multiplying the principal before one period by the above-mentioned prepayment rate to the principal and interest cash flow according to the original contracted repayment plan, By reducing the number of books, a principal cash flow that takes into account prepayments is created.

<Pull-through rate>
The risk calculation unit 215 calculates the pull-through rate for the commitment of the fixed interest rate loan as a base value by adding [1.96σ] to the average value calculated from the historical value of the user. Then, the risk calculation unit 215 calculates the cash flow by multiplying the promised repayment schedule of the commitment of the fixed interest rate loan as of the calculation reference date by this base value.

  Next, the control unit 21 of the support server 20 executes a cash flow calculation process to which interest is added (step S7-6). Specifically, the risk calculation unit 215 of the control unit 21 calculates the amount of interest income by period bucket for each position from the principal cash flow, and reflects it in the cash flow of the period bucket corresponding to the interest occurrence timing.

  Next, the control unit 21 of the support server 20 executes a current discount value calculation process (step S7-7). Specifically, the risk calculation unit 215 of the control unit 21 calculates the discount rate [exp (−R × t_k)] using the zero coupon yield (R) for each currency and each period packet. Note that t_k represents the center point of the period packet.

  Next, the risk calculator 215 multiplies the calculated discount rate by the cash flow for each period bucket to calculate the current discount value for each interest rate shock scenario.

  Next, the control unit 21 of the support server 20 executes ΔNII calculation processing (step S7-8). Specifically, the risk calculation unit 215 of the control unit 21 calculates the present discount value for the next 12 months (corresponding to six short-term period buckets).

  However, here, when the maturity comes and the interest rate is changed and replaced with the same cash flow (eg, liquid deposits and floating interest rate loans), the B / S calculation assumes that the position remains unchanged during the measurement period. To do.

  And the risk calculation part 215 calculates the total value of a period bucket as NII. Next, the risk calculation unit 215 calculates ΔNII (revenue approach) obtained by subtracting the interest level NII of the calculation reference date from the NII of the interest rate shock scenario. However, there are only two types of interest rate shock scenarios used for calculating ΔNII.

Next, the control unit 21 of the support server 20 executes an add-on calculation process for positions having interest rate options (step S7-9). Specifically, the risk calculation unit 215 of the control unit 21 allocates the principal of each position to the period bucket corresponding to the period when the interest rate option can be exercised for the add-on based on the position having the divided interest rate option property. Calculate separately from the cash flow created so far. Furthermore, the risk calculation unit 215 uses the interest rate at the time of the interest rate shock scenario set for each currency and each period packet, regards the implied volatility as an American option subject to 25%, and calculates the option value (A). calculate. The risk calculation unit 215 calculates the option value (B) in the same manner at the interest level as of the calculation reference date.
Then, the risk calculation unit 215 calculates the add-on (KAO) of the period bucket by subtracting the total of [A−B] of the buying position from the total of [A−B] of the selling position for each period bucket.

  Next, the control unit 21 of the support server 20 executes ΔEVE calculation processing (step S7-10). Specifically, the risk calculation unit 215 of the control unit 21 calculates the current discount value of all the period buckets, and then reflects the position add-on having the calculated interest rate option property. And the risk calculation part 215 makes EVE the total value of all the period buckets. Then, ΔEVE (economic value approach) is calculated by subtracting the EVE of the interest rate level at the calculation base date from the EVE of the interest rate shock scenario.

  Next, the control unit 21 of the support server 20 executes a correction duration calculation process (step S7-11). Specifically, the risk calculation unit 215 of the control unit 21 calculates the modified duration from the current discount values of all the period buckets based on the following modified duration calculation formula.

  However, CF_ (i j) is the cash flow of the period bucket j in the interest rate shock scenario i, and D_ (adji) is the modified duration in the scenario.

  Next, the control unit 21 of the support server 20 executes a result totaling process (step S7-12). Specifically, the risk calculation unit 215 of the control unit 21 aggregates the interest level at the calculation reference date, EVE, NII, ΔEVE, ΔNII, and the modified duration for each of the six interest rate shock scenarios. Then, the risk calculation unit 215 outputs the calculation result to the client terminal 10.

(Hedge position simulation)
Next, the hedge position simulation will be described with reference to FIG. In this hedge position simulation, an example of the construction of a hedge position that satisfies the conditions when the target interest rate risk-related index level and the allowable level for exceeding it, the type of derivative used, and the forecast period is set. To do.

First, the control unit 21 of the support server 20 executes a portfolio selection process (step S8-1). Specifically, the hedge simulation unit 216 of the control unit 21 outputs a selection screen for a portfolio to be simulated in the client terminal 10. Then, the hedge simulation unit 216 acquires the portfolio selected on the selection screen of the client terminal 10.
Next, the control unit 21 of the support server 20 executes a process of creating a future interest rate probability distribution (step S8-2). Specifically, the hedge simulation unit 216 of the control unit 21 creates a posterior distribution of a future zero coupon yield curve using the above-described interest rate prediction model.

Next, the control unit 21 of the support server 20 executes a discount rate creation process (step S8-3). Specifically, the hedge simulation unit 216 of the control unit 21 creates a probability distribution of the discount rate using the posterior distribution of the zero coupon / yield curve.
Next, the control part 21 of the support server 20 performs the calculation process of the interest rate risk related index (step S8-4). Specifically, the hedge simulation unit 216 of the control unit 21 uses the created probability distribution of the future interest rate, the discount rate, and the cash flow calculated in advance to calculate an interest rate risk index (ΔEVE, ΔNII, Create a target value for the modified duration distribution.

  Next, the control part 21 of the support server 20 performs the portfolio search process after the optimal hedge by optimization (step S8-5). Specifically, the hedge simulation unit 216 of the control unit 21 uses the Bayesian optimization technique to satisfy the simulation setting conditions and minimize the objective variable created from the current discount value and the complexity of the hedge (the effect is Search for a portfolio to maximize.

<Optimal hedge strategy for ΔEVE>
The optimal hedge strategy for ΔEVE is determined by solving the following optimization problems in order. Bayesian optimization is performed by successive approximation optimization using Gaussian process regression.
In order to determine the period bucket to be hedged and the interest rate risk change of the period bucket due to the hedge, a set of α_1 to α_19 that minimizes the following objective function is calculated. The meaning of the objective function is to add a penalty term created from the fluctuation range of the current discount value for each period bucket in order to avoid excessive hedging while aiming to maximize the current discount value after interest rate hedging.

The subscript i of EVE'_i indicates six types of interest rate shock scenarios.
EVE'_i is a standardization of the posterior distribution of the current discount value PV_i that takes into account the interest rate hedging effect.
α_ (i1), α_ (i2), ..., α_ (i19) are the interest rate risk change widths due to hedging for each of the 19 categories in order from the shortest term bucket (O / N). Take 10% discrete values up to 150%.
τ_ (i1), τ_ (i2),..., τ_ (i19) are present discount values for each period bucket, and h_E is a reduction amount of ΔEVE by hedge.
Derivatives used for hedging are assumed to be only plain swaps that are fixed rate and floating rate (OIS) swaps. Also, swaps between currencies are not assumed. Therefore, even if the main cause of interest rate risk is due to the basis risk between currencies, in this simulation, only the same currency (Japanese yen) is considered.
Assuming that α_ij obtained from the above optimization problem is fully hedged with an OTC plain swap that can set the assumed principal, etc. freely, the virtual position is calculated from the OIS swap rate for each period bucket and the assumed principal of the hedged item. The construction cost of is calculated.

<Optimal hedging strategy for ΔNII>
Here, the optimal hedging strategy of ΔNII is based on the optimization problem almost the same as ΔEVE, but the following objective function is minimized in order to determine the discounted cash flow of the period bucket to be hedged and the period bucket. A set of α_1 to α_6 to be converted is calculated. In addition, since the interest rate shock scenario has only two parallel shifts, it is read as such.

<Optimal hedging strategy with modified duration>
It is necessary to calculate the current discount value (PV) of the entire portfolio from the current discount value for each period bucket, and to calculate the modified duration by dividing the value calculated using the following duration calculation formula by the PV of the entire portfolio.

However, CF_ (ij) is the cash flow of the period bucket j in the interest rate shock scenario i, and D_ (adji) is the modified duration in the scenario.
The following objective function is minimized using D_ (adj i) calculated in this way. Adopting the maximum value of D_ (adj i) as the modified duration in any interest rate path, that the objective function is to minimize the error between the modified duration after hedging and the preset modified duration, as a penalty term The idea is the same as that of ΔEVE, except that the order of dispersion values is matched.

  Here, the above formula for calculating the duration is rewritten.

  Next, the control unit 21 of the support server 20 executes a process for presenting an optimal hedge strategy (step S8-6). Specifically, the hedge simulation unit 216 of the control unit 21 uses an interest rate swap (exchange of fixed interest rate and variable interest rate) to calculate a hedge position necessary for realizing an optimally hedged portfolio by optimization. .

  Next, the control unit 21 of the support server 20 executes calculation processing of the interest rate risk-related index after hedging (step S8-7). Specifically, the hedge simulation unit 216 of the control unit 21 recalculates ΔEVE, ΔNII, and the modified duration when the interest rate swap in the optimal hedging strategy is used. Then, the hedge simulation unit 216 outputs the calculation result to the client terminal 10.

As described above, according to the present embodiment, the following effects can be obtained.
(1) In the present embodiment, the core deposit prediction unit 212, the time deposit prediction unit 213, and the advanced repayment rate prediction unit 214 use a hierarchical Bayesian model. The hierarchical Bayesian model reflects changes in the economic situation from the past to the present, and Bayesian learns and optimizes the impact independently when new data (financial institution data and changes in the economic situation) is added. It is a chain-like graphical model constructed by updating.
Conventional statistical models using Bayesian statistics are constructed using static information held by financial institutions from the past, and estimation accuracy is optimized for financial institutions that have provided data. In this embodiment, by using Bayesian updating, the time series structure (dynamics) of data is taken into account, information of an arbitrary financial institution is taken in, and the estimation accuracy is optimized each time. In addition, for the operation of the conventional statistical model, the data of all explanatory variables for constructing the model is indispensable, but in this embodiment, since it is constructed by the Bayesian statistical method, even if there is missing information, missing information The risk distribution can be reasonably evaluated by using the probability distribution of the alternative. Therefore, it can cope with information of an arbitrary financial institution. Furthermore, in the conventional statistical model calculated using the individual contract information, the individual contract information needs to include all explanatory variables. However, in the present invention, even if there is missing information, the risk is reduced. Can be evaluated.

By constructing a chained graphical model with a hierarchical Bayesian structure, it is possible to appropriately reflect dramatic changes such as the economic situation thereafter, while considering information at the time of model construction as prior information.
Further, in the case of a statistical model constructed with static data, the estimation accuracy deteriorates with the passage of time after construction. In this embodiment, since the statistical model is updated using only newly added data, the statistical model can be updated efficiently. This makes it possible to quickly optimize the environment that was not assumed at the time of model construction and maintain high estimation accuracy while taking into account past data transitions.

(2) In this embodiment, by using a hierarchical Bayesian structure, all the variables constituting the statistical model have a probability distribution. Therefore, even if you do not know the data structure, it is possible to make a reasonable estimate by updating using the probability distribution of a financial institution that uses a probability distribution created in advance or using the prior probability distribution as it is. . Furthermore, by adding an adjustment term that expresses the variation in characteristics of each financial institution, it is possible to make an estimation that reflects the individuality of the financial institution, such as regional characteristics and management policy differences.
In addition, since the influence of new data can be reflected by Bayesian update, deterioration of the performance of the statistical model can be avoided while maintaining the characteristics of the model before the update to some extent (while maintaining the continuity of the estimation result of the model).

  (3) In the present embodiment, the control unit 21 of the support server 20 executes the acquisition process of the sum total of the principal, the number of contracts, the average value, the standard deviation, and the quantile summary data for each data item (step S6- 1). And the control part 21 of the assistance server 20 performs the estimation process of the parameter of an advance repayment rate estimation model (step S6-2). The calculation can be performed without using the contract details of individual bonds while suppressing the error by calculating the summary instead of calculating the cash flow for each bond.

In addition, you may change the said embodiment as follows.
In the above embodiment, the core deposit prediction unit 212 to the advanced repayment rate prediction unit 214 use a hierarchical Bayesian model. If at least a part uses a hierarchical Bayesian model, other statistical models can be used.
In the above embodiment, the core deposit balance, fixed deposit amount, and advanced repayment rate are calculated. The information is not limited to these as long as it is necessary for calculating the risk calculation.
In the above embodiment, the client terminal 10 and the support server 20 are used for risk evaluation. The hardware configuration is not limited to these.

  DESCRIPTION OF SYMBOLS 10 ... Client terminal, 20 ... Risk evaluation system, 21 ... Control part, 211 ... Interest rate prediction part, 212 ... Core deposit prediction part, 213 ... Time deposit prediction part, 214 ... Prepayment rate prediction part, 215 ... Risk calculation part, 216 ... Hedge simulation unit, 22 ... Public information storage unit, 23 ... Individual information storage unit, 32 ... Interest rate prediction model.

Claims (4)

A public information storage unit for storing public nationwide information;
An individual information storage unit for storing individual summary information representing an index in a specific organization;
A risk evaluation system comprising a control unit for calculating a risk index,
The control unit is
In order to calculate interest rates and institutional indicators using publicly available national information, an interest rate prediction model and indicator prediction model consisting of a chained graphical model having a hierarchical Bayesian structure is generated,
When obtaining individual summary information for calculating an engine index from a specific institution, using the individual summary information, adjusting the index prediction model,
According to the interest rate calculated based on the interest rate prediction model, the engine index prediction value is calculated using the adjusted index prediction model, the risk index is calculated based on the engine index prediction value,
A risk evaluation system characterized by performing Bayesian updating on the index prediction model using newly disclosed national information or newly acquired individual summary information.   The risk evaluation system according to claim 1, wherein the institutional index includes a core deposit balance, a fixed deposit balance, and an advanced repayment rate. A public information storage unit for storing public nationwide information;
An individual information storage unit for storing individual summary information representing an index in a specific organization;
A method for evaluating interest rate risk using a risk evaluation system including a control unit for calculating a risk index,
The control unit is
In order to calculate interest rates and institutional indicators using publicly available national information, an interest rate prediction model and indicator prediction model consisting of a chained graphical model having a hierarchical Bayesian structure is generated,
When obtaining individual summary information for calculating an engine index from a specific institution, using the individual summary information, adjusting the index prediction model,
According to the interest rate calculated based on the interest rate prediction model, the engine index prediction value is calculated using the adjusted index prediction model, the risk index is calculated based on the engine index prediction value,
A risk evaluation method comprising performing Bayesian update on the index prediction model using newly disclosed national information or newly acquired individual summary information. A public information storage unit for storing public nationwide information;
An individual information storage unit for storing individual summary information representing an index in a specific organization;
A program for evaluating interest rate risk using a risk evaluation system including a control unit for calculating a risk index,
The control unit
In order to calculate interest rates and institutional indicators using publicly available national information, an interest rate prediction model and indicator prediction model consisting of a chained graphical model having a hierarchical Bayesian structure is generated,
When obtaining individual summary information for calculating an engine index from a specific institution, using the individual summary information, adjusting the index prediction model,
According to the interest rate calculated based on the interest rate prediction model, the engine index prediction value is calculated using the adjusted index prediction model, the risk index is calculated based on the engine index prediction value,
A risk evaluation program that functions as a means for performing Bayesian updating on the index prediction model using newly disclosed national information or newly acquired individual summary information.