JPH10222488A - Model supply system for risk management method of monetary property - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce burden on trial and error when the system of VaR calculation is newly introduced by inputting a parameter and a setting condition, which are necessary at the time of calculating a value at risk (VaR) and calculating plural piece of sensitive degree data on propoerty from possessed property data. SOLUTION: When market data 13, possessed property data 14 and a user setting condition 15 are inputted, a VaR calculation processor 11 outputs VaR data 16 where a maximum predicted loss that property containing monetary derivative products such as futures and options, which monetary facilities possess, receive at constant probability during a possessing period is statistically displayed. When real gain and loss data 17 and VaR data 16 are inputted, a verification processor 12 outputs verification result data 18. Thus, models for AvR calculation peculiar to the monetary facilities is automatically selected only by setting the parameter which requires examination as an input condition.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a financial institution or
With regard to the risk management of financial assets held by corporate entities, when developing a model for the maximum expected loss represented by the value-at-risk of assets, including financial derivatives,
Area related to how to provide models that are compatible with the market environment and the composition of assets held, which are factors that change at risk.

[0002]

2. Description of the Related Art Conventionally, the risk management of assets held by financial institutions, including financial derivatives, has been performed as follows. First,
As discussed in “Calculation of Value-at-Risk and Risk / Return Simulation” (Bank of Japan Monthly Report, 1995.4) (P13-18), the holding period called Value-at-Risk (VaR) Calculates a risk index that statistically displays the maximum expected loss that can occur in the portfolio and compares it with the expected return of the assets held and the capital of the financial institution, thereby Judged the validity.

[0003]

However, the conventional method has the following problems.

[0004] There are various variations in the method of calculating VaR. As discussed in “Portfolio Risk Management of Financial Institutions-Focusing on Value-at-Risk” (Securities Analyst Journal 1995.8) (P12-21), users (financial institution risk managers) ) Determines the method of calculating volatility and correlation coefficient from the setting of parameters, and determines a model unique to a financial institution. The model depends not only on the market environment but also on the composition of the portfolio held by the financial institution, and the model differs for each financial institution.

[0005] Therefore, in a financial institution that newly introduces a VaR calculation system, a risk management person calculates VaR from data for many years using a single model, compares it with actual profit and loss, and actually calculates VaR. If not, it is necessary to calculate VaR again using a different model, and iteratively perform trial and error until a model that is actually fit is created, which poses a problem that the risk management staff takes time and physical burden.

In order to solve the above problems, the present invention provides a financial institution that newly introduces a VaR calculation system with a model that is considered optimal by reducing the burden of risk management personnel involved in trial and error. With the goal.

[0007]

[MEANS FOR SOLVING THE PROBLEMS] In order to analyze the risk of an asset containing a financial derivative, the value of the asset, which is an indicator for statistically indicating the maximum expected loss that the asset incurs at a certain probability during the holding period, is displayed. A model providing system for risk analysis of financial assets that provides an at-risk model, comprising: means for inputting parameters necessary for value-at-risk calculation and setting conditions for a data processing method; and Means for generating a plurality of value-at-risk models by calculating a plurality of volatility data and correlation coefficient data from observed data and calculating a plurality of sensitivity data of the assets from owned asset data according to the set conditions. And comparing the generated models with the actual profit or loss after a certain holding period of the asset to determine whether the model actually fits Means for performing a verification can be achieved by having a means for providing by selecting one model from the plurality of models based on the verification result.

[0008]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an apparatus configuration of the present invention.

[0010] Market data 13 refers to market data necessary for VaR calculation. For example, for interest rate products, interest rate data such as LIBOR (one month) and swap rate (two-year 1.61), for exchange products, exchange rate data for various currencies such as the dollar / yen exchange rate (107.52), and for stock products , Refers to stock price and stock index data such as the stock index Nikkei 225 (22,292.91), and in the case of option products, refers to option underlying asset price data such as the short-term interest rate futures price (98.65).

The owned asset data 14 refers to asset data including financial derivatives such as futures and options held by the financial institution. For example, in the case of JGB futures (type: JGB futures, brand code: super long term, contract month: 3, trading category: sell, contract price: 119.00, quantity: 10, delivery date: 3 /
20, coupon: 6, interest payment date: 9/20, 3/20 ..., transaction unit: 100,000,000, maturity: 20). The actual profit / loss data 17 includes actual profit / loss data (for example, 24,562,300) generated by buying and selling assets including financial derivatives such as futures and options held by financial institutions.
Circles).

[0012] The VaR data 16 is a VaR calculation process in which the maximum expected loss that the asset including financial derivatives such as futures and options held by the financial institution incurs with a certain probability during the holding period is statistically displayed. Refers to value data (for example, 24,785,500 yen).

The verification result data 18 includes actual profit and loss data 17.
And VaR data 16 are described in the "Supervisory Framework for Using Backtesting in an Internal Model Approach to Calculating Capital Required for Market Risk" (Basel Committee on Banking Supervision, January 1996). This is data that records the number of times the value of the actual profit / loss data 17 exceeds the value of the VaR data 16 by comparing according to the verification method, and as shown in the table of FIG. 4, the setting conditions of the user of the VaR model to be verified. 41, the number of measurement days 42 in the VaR calculation period, the VaR value 45 and the actual profit / loss 46 at that time, the number of excesses 43, and the ranking 44 of the model considered to be optimal.

When the market data 13, the owned asset data 14, and the user setting conditions 15 are input, the VaR calculation processing device 11 allows assets including financial derivatives such as futures and options held by the financial institution to have a certain probability during the holding period. VaR data 16 that statistically displays the maximum expected loss amount incurred
Is output.

Upon inputting the actual profit / loss data 17 and the VaR data 16, the verification processing device 12 sets “the supervisory framework for using backtesting in the internal model approach used for calculating the required capital amount for market risk”. (Basel Committee on Banking Supervision 1996 1
Verification result data 1 according to the verification method described in
8 is output.

The processing flow of this embodiment will be described with reference to FIGS.

First, a risk manager in charge of the user inputs the following setting conditions in the user setting condition input 21 of FIG.

[0018] 1. Data observation period 2. 2. How to supplement data 3. Whether or not singular values / trends are removed. 4. Calculation method of volatility / correlation coefficient Sensitivity measurement method VaR calculation target period 31 in FIG. 3 (for example, 95/04/01 to 95/09/30 as 41 in FIG. 4), data observation period 32 (for example, 94/94 as 41 in FIG. 4) 04/01 to 95/09/30), and select from methods such as linear interpolation spline interpolation that does not complement data interpolation method 33. (E.g., linear or spline as indicated by 41 in FIG. 4) Singular value, trend removal presence / absence 34 is selected from not performed.

The volatility / correlation coefficient calculation method 35 is selected from methods such as daily calculation, 10-day intervals, 20-day intervals, and 30-day intervals.

From the sensitivity measurement method 36, 1 bp −
Select from methods such as averaging 1 bp difference absolute values on the side.

When there is an item for which the user does not select a setting condition, a predetermined condition is used. Further, since the form of the input data and the output data does not change in the selection of the setting condition, the selection of the setting condition does not affect the entire processing flow.

Next, a plurality of VaRs are calculated in the following process for all combinations considered from the selected parameters and methods.

First, singular values and trends are removed from the observed data, and volatility data is calculated from the data after the data is complemented. Subsequently, correlation coefficient data is calculated from the volatility data. On the other hand, market value evaluation data of the held asset is calculated from the held asset data, and sensitivity data is calculated from the market value data. Next, VaR data is calculated from the volatility data, the correlation coefficient data, and the sensitivity data.

A table shown in FIG. 4 is created from the output VaR data and the actual profit / loss data.

The comparative verification of the value of VaR and the actual profit or loss is described in "Supervisory Framework for Using Backtesting in an Internal Model Approach Used to Calculate the Capital Amount Required for Market Risk" (Basl Committee on Banking Supervision, 1996). In accordance with the verification method described in (January), the number of times that the value of the actual profit and loss data exceeds the value of the output VaR data is measured, and the models with the least number of times of excess are set as the models that are closer to the optimal one.

The output to the screen is shown in FIG. 4, FIG. 5, and FIG. FIG. 4 shows the generated model (,...) And the actual profit and loss on the vertical axis, and the setting condition 4 for each of the generated models.
1 is a table showing the verification results with the calculated VaR 45, the number of days 42 in which VaR was measured, the number of times VaR exceeded the actual profit and loss 43, and the rank 44 of the optimal model on the horizontal axis.

FIG. 5 is a graph showing the time series (the horizontal axis represents the date and the vertical axis represents the VaR value and the actual profit and loss) for each of the generated models. Since the value of VaR is the maximum expected loss, it is indicated by a negative value. When the actual profit / loss amount falls below the value of VaR, it is counted as an excess number.

FIG. 6 shows an arrangement of the verification result table of FIG.
The edited version is displayed in the order of the most suitable model. By comparing the setting conditions, the tendency of each selection method can be considered.

[0029]

As described above, the risk management person has conventionally constructed a model with a time and physical burden by repeating trial and error, but in the present invention, input the parameters that need to be examined. A model for VaR calculation unique to a financial institution can be automatically selected only by setting as a condition.

Further, the risk management person can grasp the characteristics and trends of each method by performing a trial calculation of many methods, and can change the model according to changes in the market environment after operating the system for calculating the VaR. It becomes possible to respond flexibly.

[Brief description of the drawings]

FIG. 1A is a configuration diagram of an apparatus for carrying out the present invention.
(B) A device configuration diagram for implementing the present invention.

FIG. 2 is a processing flowchart showing an embodiment according to the present invention.

FIG. 3 is a diagram showing an image of a parameter condition input screen of a risk manager in charge as a user.

FIG. 4 is a diagram of a model verification result table.

FIG. 5 is a diagram showing a screen image of an output graph.

FIG. 6 is a diagram showing a screen image of an output table.

[Explanation of symbols]

11: VaR calculation processing, 12: comparison verification processing, 13: market data, 14: owned asset data, 15: user setting condition input, 16: VaR data, 17: actual profit and loss data, 18: verification result data

Claims (3)

[Claims] Claims: 1. To analyze the risk of an asset, including a derivative, a value, which is an index that statistically indicates the maximum expected loss that the asset incurs with a certain probability during the holding period.
A model providing system for risk analysis of financial assets that provides an at-risk model, a means for inputting parameters and setting conditions for a data processing method required for value-at-risk calculation,
Generate multiple value-at-risk models by calculating a plurality of volatility data and correlation coefficient data from observation data and calculating a plurality of sensitivity data of the assets from owned asset data according to the input setting conditions Means for comparing the generated plurality of models with actual profit or loss after a certain holding period of the asset to verify whether the model actually fits, and the plurality of models based on the verification result. Means for selecting and providing one model from among the models described above. 2. The value-at-risk calculation process, which is an index that statistically displays the maximum expected loss amount incurred at a certain probability according to claim 1, automatically when the user does not select a parameter set by the user. A model providing system in a risk analysis of a financial asset, characterized by having a process of setting parameters dynamically. 3. Comparing all calculated values of claim 1 with actual profits and losses after a certain holding period of an asset including a financial derivative, and verifying whether the model actually fits, A model providing system for risk analysis of financial assets, characterized in that characteristics and trends of each method can be grasped from verification results.