Method and system for predicting expected credit loss at future time point

Abstract

The invention discloses a prediction method and system for expected credit loss at a future time point, and the method comprises the steps: S1, building a new prediction scheme which comprises a legal person mechanism, a value reduction scheme, a prediction dimension, a business data date and a prediction date; s2, preparing business data required by prediction, wherein the business data comprises stock business data of business data dates, stock asset five-level classification modification data, stock asset subject rating modification data, verification data and incremental business data; s3, according to a preset rule and the business data, generating total business data of the prediction date; s4, according to the total business data of the prediction date, the value reduction scheme, the prediction date and value reduction calculation, obtaining a value reduction result of the prediction date; and S5, querying the value reduction result of the prediction date and comparison information of the value reduction result of the prediction date and the value reduction result of the business data date. According to the method, the influence of future time point expected credit loss on related financial, supervision and internal assessment indexes is evaluated in time, and the problem of pain points of financial institutions in the aspect of value reduction measurement is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of financial systems, and in particular to a method and system for predicting expected credit losses at future points in time. Background Art

[0002] The expected credit losses of financial assets are related to many financial, regulatory and internal assessment indicators of financial institutions. The current expected credit loss measurement is mainly carried out at the current point in time. Based on the business data at the current point in time, the credit loss of each exposure data is measured, and the impairment calculation of each scheme corresponding to the exposure is performed according to the scheme configuration corresponding to each exposure. The configuration and calculation are based on existing data and are not predictive and forward-looking.

[0003] For the measurement of expected credit losses at future points, business personnel usually use the impairment results of the current business date data as the basis to manually calculate the impairment results of the forecast date change data, and add the two to calculate the impairment measurement results of all exposures on the forecast date, which has certain accuracy issues. Manual operations by business personnel have certain requirements on their work ability and work experience, so the workload and work difficulty of the corresponding personnel need to be shared by the system.

[0004] Therefore, it has become an urgent need for financial institutions to establish an objective technical solution for measuring expected credit losses at future points in time and to timely evaluate the impact of expected credit loss results on financial, regulatory, and internal assessment indicators. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for predicting expected credit losses at future points in time, which can timely evaluate the impact of expected credit losses at future points in time on relevant financial, regulatory, and internal assessment indicators, and solve the pain points of financial institutions in impairment measurement.

[0006] The present invention provides a method for predicting expected credit losses at a future point in time, the method comprising:

[0007] S1. Create a new forecasting plan, which includes: legal entity, impairment plan, forecasting dimension, business data date and forecasting date;

[0008] S2. Prepare the business data required for the forecast, the business data including: stock business data on the business data date, stock asset five-level classification modification data, stock asset subject rating modification data, write-off data and incremental business data;

[0009] S3. Generate the full amount of business data for the forecast date according to the preset rules and the business data;

[0010] S4. Based on the full amount of business data on the forecast date, the impairment plan and the forecast date, the impairment calculation function is called to calculate the impairment result on the forecast date;

[0011] S5. Query the impairment result of the forecast date and compare it with the impairment result of the business data date.

[0012] Preferably, the stock business data on the business data date includes: impairment risk exposure business data of corporate loans, retail loans, corporate off-balance sheet, credit cards, interbank business, bond investments, non-standard investments, and bill business;

[0013] The five-level classification modification data of stock assets includes: new five-level classification information of the stock business data to be adjusted, including: legal entity, forecast date, five-level classification, and asset number;

[0014] The existing asset subject rating modification data includes: new subject rating information of the existing business data to be adjusted, including: legal entity, forecast date, subject rating, asset number, name of transaction subject / issuer / financier / accepting bank;

[0015] The write-off data is data information of the assets to be written off, including: legal entity, write-off date, and asset number;

[0016] The incremental business data refers to the new business data added on the forecast date.

[0017] Preferably, step S3 comprises:

[0018] S31, updating the stock business data in the business data according to a preset rule, including:

[0019] Update overdue days: Calculate the time interval between the business data date and the forecast date, assuming it is n days. If the current debt overdue days is 0, then the debt overdue days on the forecast date will still be 0. If the current debt overdue days is greater than 0, then the debt overdue days on the forecast date will be the original overdue days + n.

[0020] Update risk classification: modify data according to the five-level classification of stock assets and update the five-level classification information of related debt items;

[0021] Update the subject rating: modify the data based on the subject rating of the existing assets and update the subject rating information of the relevant debt items;

[0022] Update write-off information: update the write-off information of related debt items according to the write-off data;

[0023] S32: superimpose the updated existing business data with the incremental business data to generate full business data for the forecast date.

[0024] Preferably, step S4 comprises:

[0025] S41, based on the risk classification, overdue days, and write-off information of the forecast date, according to the stage division rules in the impairment plan, recalculate the customer-level risk classification and overdue days to determine the impairment stage of the forecast date;

[0026] S42. According to the impairment stage of the forecast date, different impairment parameters are called to calculate a preliminary impairment result;

[0027] S43. According to the weights of optimism, benchmark and pessimism, the preliminary impairment results are weighted and summed to calculate the final impairment result of the forecast date.

[0028] Preferably, the impairment stage of the forecast date includes stage one, stage two and stage three, and step S42 includes:

[0029] If the impairment stage of the forecast date is stage one, the preliminary impairment result is the expected credit loss for the next 12 months, ECL = PD × LGD × EAD, where ECL is the preliminary impairment result, PD is the forward-looking adjusted one-year probability of default, LGD is the default loss of the corresponding guarantee method, and EAD is the default risk exposure at the current measurement point;

[0030] If the impairment stage at the forecast date is stage 2, the preliminary impairment result is the expected credit loss during the duration; if the current exposure method is used, ECL = EAD × CPD i ×LGD, where i is the duration, CPD i is the cumulative default probability after forward-looking adjustment in the i-th period; if the exposure method is used in each period, ECL = Σ N i＝1 (EAD i ×MPD i ×LGD i ) / (1+Rate) (i-1) , where EAD i is the default risk exposure of the i-th period, MPD i is the marginal default probability of the i-th period, LGD i is the default loss of the i-th period, Rate is the loan interest rate, i and N are positive integers;

[0031] If the impairment stage on the forecast date is stage three, the preliminary impairment result is the expected credit loss during the life of the loan, ECL = EAD × LGD.

[0032] Preferably, the impairment result of the forecast date and its comparison information with the impairment result of the business data date include: forecast type code, forecast type name, legal entity, asset number, risk exposure / grouping, original five-level classification, original entity rating, original overdue days, original stage judgment result, forecast overdue days, forecast stage result, forecast impairment result, impairment supplementary provision amount, and impairment change amount.

[0033] Based on the same inventive concept, the present invention also provides a prediction system for expected credit losses at future points in time, which is used to implement the prediction method. The prediction system includes:

[0034] The impairment forecast module is used to create a new forecast plan, which includes: legal entity, impairment plan, forecast dimension, business data date and forecast date;

[0035] A data processing module, used to prepare business data required for prediction, the business data including: stock business data on the business data date, stock asset five-level classification modification data, stock asset subject rating modification data, write-off data and incremental business data;

[0036] The impairment prediction module is further used to generate the full amount of business data on the forecast date according to the preset rules and the business data; and to call the impairment calculation function according to the full amount of business data on the forecast date, the impairment plan and the forecast date to calculate the impairment result on the forecast date;

[0037] The statistical query module is used to query the impairment result of the forecast date and its comparison information with the impairment result of the business data date.

[0038] Preferably, the stock business data on the business data date includes: impairment risk exposure business data of corporate loans, retail loans, corporate off-balance sheet, credit cards, interbank business, bond investments, non-standard investments, and bill business;

[0039] The five-level classification modification data of stock assets includes: new five-level classification information of the stock business data to be adjusted, including: legal entity, forecast date, five-level classification, and asset number;

[0040] The existing asset subject rating modification data includes: new subject rating information of the existing business data to be adjusted, including: legal entity, forecast date, subject rating, asset number, name of transaction subject / issuer / financier / accepting bank;

[0041] The write-off data is data information of the assets to be written off, including: legal entity, write-off date, and asset number;

[0042] The incremental business data refers to the new business data added on the forecast date.

[0043] Preferably, the impairment prediction module is specifically used for:

[0044] Updating the stock business data in the business data according to the preset rules includes:

[0045] Update overdue days: Calculate the time interval between the business data date and the forecast date, assuming it is n days. If the current debt overdue days is 0, then the debt overdue days on the forecast date will still be 0. If the current debt overdue days is greater than 0, then the debt overdue days on the forecast date will be the original overdue days + n.

[0046] Update risk classification: modify data according to the five-level classification of stock assets and update the five-level classification information of related debt items;

[0047] Update the subject rating: modify the data based on the subject rating of the existing assets and update the subject rating information of the relevant debt items;

[0048] Update write-off information: update the write-off information of related debt items according to the write-off data;

[0049] The updated existing business data is superimposed on the incremental business data to generate the full business data for the forecast date;

[0050] The impairment prediction module is also specifically used for:

[0051] Based on the risk classification, overdue days, and write-off information on the forecast date, according to the stage division rules in the impairment plan, recalculate the customer-level risk classification and overdue days to determine the impairment stage on the forecast date;

[0052] According to the impairment stage of the forecast date, different impairment parameters are called to calculate the preliminary impairment results;

[0053] The preliminary impairment results are weighted and summed according to the optimistic, baseline and pessimistic weights to calculate the final impairment result on the forecast date.

[0054] Preferably, the system further comprises:

[0055] The impairment parameter module is used to set the rules of impairment parameters and perform corresponding calculations. The calculation results are directly configured and called by the impairment scheme link. The impairment parameters include the following parameters: risk grouping, stage division, PD migration matrix, PD main scale, LGD management, prospective management, EAD management and CCF management. Each parameter sets the corresponding rules and performs measurement;

[0056] The impairment scheme management module is used to select the risk exposure involved in the impairment scheme. It supports adding risk groups under risk exposure, and then flexibly configures the impairment parameters used under the risk exposure / group and sets the priority. Under normal circumstances, a risk group can use the impairment parameters of the risk group or the impairment parameters of the risk exposure to which it belongs, and cannot use the impairment parameters of other risk groups under the exposure to which it belongs.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] In the prior art, for the measurement of expected credit losses at future points, business personnel usually manually calculate the impairment results of the change data on the forecast date based on the impairment results of the current business date data, and add the two to estimate the impairment measurement results of all exposures on the forecast date, which has certain accuracy problems. The present invention, however, generates the full amount of data required for the forecast by analyzing the current business data, analyzing and processing the existing business data and the incremental data entered by the business personnel at the corresponding forecast time point. The specific analysis and processing is mainly through risk classification, subject rating, overdue days, write-off status and other multi-level and multi-dimensional parameters and business attribute changes, and also supports business personnel to supplement incremental business data. The above multi-dimensional processing generates the full amount of business data on the forecast date, paving the way for subsequent impairment forecast calculation and other processes.

[0059] The present invention uses multi-level and multi-dimensional business parameters such as risk classification, entity rating, overdue days, write-off status, and different business settings of major exposure businesses to simulate the impairment measurement process at future forecast points, and uses the actual measurement process as a basis for additional expansion. It can perform multiple impairment forecast calculations with different configurations, different parameters, different exposures, and different incremental data, thereby providing business personnel with a variety of impairment measurement results for reference and prediction of future measurement data, meeting the needs of multi-dimensional consideration and prediction of the business. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of a flow chart of a method for predicting expected credit losses at a future point in time provided by the present invention;

[0061] Figure 2 A schematic diagram of a flow chart of a method for predicting expected credit losses at a future point in time provided by the present invention;

[0062] Figure 3 A schematic diagram of a sub-process of a method for predicting expected credit losses at a future point in time provided by the present invention;

[0063] Figure 4 A schematic diagram of a sub-process of a method for predicting expected credit losses at a future point in time provided by the present invention;

[0064] Figure 5A schematic diagram of the structure of a prediction system for expected credit losses at future points in time provided by the present invention;

[0065] Figure 6 A schematic diagram of the workflow of a system for predicting expected credit losses at future points in time provided by the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0068] like Figure 1-4 As shown, the present invention provides a method for predicting expected credit losses at a future point in time, the prediction method comprising:

[0069] S1. Create a new forecast plan, which includes: legal entity, impairment plan, forecast dimension, business data date and forecast date;

[0070] Assuming that the current business data date is 2015.1.20, a new impairment forecast plan for the future point in time can be created. The key elements include: legal entity - XXX Bank, impairment plan - A impairment calculation plan, business data date - 2024.12.31, forecast date - 2025.2.28, forecast dimensions - risk classification & entity rating & write-off & overdue days.

[0071] S2. Prepare the business data required for forecasting, including: stock business data on the business data date, stock asset five-level classification modification data, stock asset subject rating modification data, write-off data and incremental business data;

[0072] The stock business data on the business data date include: impairment risk exposure business data of corporate loans, retail loans, corporate off-balance sheet, credit cards, interbank business, bond investments, non-standard investments, and bill business.

[0073] The five-level classification modification data of stock assets includes: the new five-level classification information of the stock business data to be adjusted, including: legal entity, forecast date, five-level classification, asset number. If the five-level classification of certain exposure business data needs to be adjusted, the new five-level classification information of the stock business data to be adjusted needs to be imported, and the main elements include: legal entity, forecast date, five-level classification, asset number, etc.

[0074] The modification data of the subject rating of existing assets includes: the new subject rating information of the existing business data to be adjusted, including: legal entity, forecast date, subject rating, asset number, name of the transaction subject / issuer / financier / accepting bank. If the subject rating of certain exposure business data needs to be adjusted, the new subject rating information of the existing business data to be adjusted needs to be imported, and the main elements include: legal entity, forecast date, subject rating, asset number, name of the transaction subject / issuer / financier / accepting bank, etc.

[0075] Write-off data is the data information of the assets to be written off, including: legal entity, write-off date, asset number. If you need to write off certain assets, you need to import the write-off data information, the main elements of which include: legal entity, write-off date, asset number, etc.

[0076] Incremental business data refers to the newly added business data on the forecast date. If new business data is required on the forecast date, the incremental data information on the forecast date needs to be imported, and the business data element information is imported according to different exposures. The import of new data for multiple exposures is supported.

[0077] S3. Generate full business data for the forecast date based on preset rules and business data;

[0078] S31. Updating the stock business data in the business data according to the preset rules, including:

[0079] Update overdue days: Calculate the time interval between the business data date and the forecast date, assuming it is n days. If the current debt overdue days is 0, then the debt overdue days on the forecast date will still be 0. If the current debt overdue days is greater than 0, then the debt overdue days on the forecast date will be the original overdue days + n.

[0080] Update risk classification: Modify data based on the five-level classification of existing assets and update the five-level classification information of related debt items.

[0081] Update the subject rating: modify the data according to the subject rating of existing assets and update the subject rating information of related bonds.

[0082] Update write-off information: Update the write-off information of related debts based on the write-off data.

[0083] S32. Superimpose the updated existing business data with the incremental business data to generate the full business data for the forecast date.

[0084] S4. Based on the full business data, impairment plan and forecast date on the forecast date, the impairment calculation function is called to calculate the impairment result on the forecast date;

[0085] S41. Based on the risk classification, overdue days, and write-off information on the forecast date, the customer-level risk classification and overdue days are recalculated according to the stage division rules in the impairment plan to determine the impairment stage on the forecast date;

[0086] S42. According to the impairment stage of the forecast date, different impairment parameters are called to calculate the preliminary impairment result;

[0087] The impairment stage on the forecast date includes stage 1, stage 2 and stage 3. For example, if the original risk classification of debt item A is normal and the revised risk classification is special mention, the impairment stage on the forecast date is stage 2, not stage 1 on the original business date; if the original overdue days of debt item B is 35 and the overdue days on the forecast date is 94, the impairment stage on the forecast date is stage 3, not stage 2 on the original business date.

[0088] Step S42 includes:

[0089] If the impairment stage on the forecast date is stage one, the preliminary impairment result is the expected credit loss for the next 12 months, ECL = PD × LGD × EAD, where ECL is the preliminary impairment result, PD is the forward-looking adjusted one-year probability of default, LGD is the default loss of the corresponding guarantee method, and EAD is the default risk exposure at the current measurement time;

[0090] If the impairment stage on the forecast date is stage 2, the preliminary impairment result is the expected credit loss during the life of the asset; if the current exposure method is used, ECL = EAD × CPD i ×LGD, where i is the duration, CPD i is the cumulative default probability after forward-looking adjustment in the i-th period (if the duration is 3 years, CPD3 is used); if the exposure method is used, ECL = Σ N i＝1 (EAD i ×MPD i ×LGD i ) / (1+Rate) (i-1) , where EAD i is the default risk exposure of the i-th period, MPD i is the marginal default probability of the i-th period, LGD i is the default loss of the i-th period, Rate is the loan interest rate, i and N are positive integers;

[0091] If the impairment stage on the forecast date is stage three, the preliminary impairment result is the expected credit loss during the life of the asset, because the PD is 100%, and ECL = EAD × LGD.

[0092] S43. Based on the optimistic, benchmark and pessimistic weights, the preliminary impairment results are weighted and summed to calculate the final impairment result on the forecast date.

[0093] ECL 调整后 =ECL 乐观 *Weight 乐观 +ECL 基准 *Weight 基准 +ECL 悲观 *Weight 悲观

[0094] S5. Query the impairment result of the forecast date and compare it with the impairment result of the business data date.

[0095] The impairment results on the forecast date and their comparison with the impairment results on the business data date include: forecast type code, forecast type name, legal entity, asset number, risk exposure / grouping, original five-level classification, original entity rating, original overdue days, original stage judgment results, forecast overdue days, forecast stage results, forecast impairment results, impairment supplementary amount, impairment change amount, etc.

[0096] like Figure 5 As shown, an embodiment of the present invention further provides a prediction system for expected credit losses at future points in time, including:

[0097] The impairment prediction module 100 is used to create a new prediction plan, which includes: legal entity, impairment plan, prediction dimension, business data date and prediction date;

[0098] The data processing module 200 is used to prepare the business data required for the forecast, and the business data includes: the stock business data of the business data date, the five-level classification modification data of the stock assets, the rating modification data of the stock assets subject, the write-off data and the incremental business data;

[0099] The impairment prediction module 100 is further used to generate full business data on the forecast date according to preset rules and business data; based on the full business data on the forecast date, the impairment plan and the forecast date, the impairment calculation function is called to calculate the impairment result on the forecast date;

[0100] The statistical query module 300 is used to query the impairment result of the forecast date and compare it with the impairment result of the business data date.

[0101] like Figure 6As shown, the prediction system in the embodiment of the present invention also includes: an impairment parameter module and an impairment scheme management module. First, the impairment prediction module 100 creates a new prediction scheme; the data processing module 200 receives / imports stock business data, five-level classification / subject rating modification data, write-off data, and incremental business data; the impairment parameter module sets the rules for impairment parameters and performs corresponding calculations, and the calculation results are directly configured and called for the impairment scheme link. The parameters involved mainly include: risk grouping, stage division, PD migration matrix, PD main scale, LGD management, forward-looking management, EAD management, CCF management, etc., and each parameter sets the corresponding rules and performs measurement; the impairment scheme management module can select the risk exposure involved in the impairment scheme, and supports adding risk groups under risk exposure, and then flexibly configures the impairment parameters used under risk exposure / grouping and sets priorities. Normally, a risk group can use the impairment parameters of the risk group or the impairment parameters of the risk exposure to which it belongs, but cannot use the impairment parameters of other risk groups under the exposure to which it belongs; the impairment prediction module generates the full business data on the forecast date based on the data processing module 200, and calculates and measures the impairment in combination with the forecast plan and the full business data on the forecast date. There is no limit on the number of impairment calculations, and business personnel can calculate and debug impairment plans, parameter plans, etc. multiple times according to actual needs; the statistical query module 300 queries the detailed results of the impairment prediction plan and compares it with the impairment results on the business data date.

[0102] Compared with the prior art, the advantages of the present invention are:

[0103] The existing system mainly measures the impairment of existing business data, and has no clear results for future impairment data changes, or only incrementally enters business data for impairment calculation as a reference for forecasting information. Relatively speaking, the present invention, through the setting of multi-dimensional prediction rules, can generate the full amount of business data on the forecast date based on existing data and incremental business data, and then can perform impairment forecast calculations for the entire exposure, rather than only predicting the data changes, so that the forecast results are more accurate.

[0104] The impairment prediction function of this system can use different parameter configurations to make multi-dimensional and multiple parameter adjustments and predictions based on the business data generated by the above steps, according to the parameter configuration of business personnel and different business data input. It can view the prediction results at future points with multiple prediction schemes, and effectively evaluate the impact of impairment results on financial indicators, regulatory indicators and internal assessment indicators.

[0105] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting expected credit losses at a future point in time, characterized in that: The prediction method comprises: S1. Create a new forecasting plan, which includes: legal entity, impairment plan, forecasting dimension, business data date and forecasting date; S2. Prepare the business data required for the forecast, the business data including: stock business data on the business data date, stock asset five-level classification modification data, stock asset subject rating modification data, write-off data and incremental business data; S3. Generate the full amount of business data for the forecast date according to the preset rules and the business data; S4. Based on the full amount of business data on the forecast date, the impairment plan and the forecast date, the impairment calculation function is called to calculate the impairment result on the forecast date; S5. Query the impairment result of the forecast date and compare it with the impairment result of the business data date.

2. The method for predicting expected credit losses at future points in time according to claim 1, characterized in that: The stock business data on the business data date include: impairment risk exposure business data of corporate loans, retail loans, corporate off-balance sheet, credit cards, interbank business, bond investments, non-standard investments, and bill business; The five-level classification modification data of stock assets includes: new five-level classification information of the stock business data to be adjusted, including: legal entity, forecast date, five-level classification, and asset number; The existing asset subject rating modification data includes: new subject rating information of the existing business data to be adjusted, including: legal entity, forecast date, subject rating, asset number, name of transaction subject / issuer / financier / accepting bank; The write-off data is data information of the assets to be written off, including: legal entity, write-off date, and asset number; The incremental business data refers to the new business data added on the forecast date.

3. The method for predicting expected credit losses at future points in time according to claim 2, characterized in that: Step S3 includes: S31, updating the stock business data in the business data according to a preset rule, including: Update overdue days: Calculate the time interval between the business data date and the forecast date, assuming it is n days. If the current debt overdue days is 0, then the debt overdue days on the forecast date will still be 0. If the current debt overdue days is greater than 0, then the debt overdue days on the forecast date will be the original overdue days + n. Update risk classification: modify data according to the five-level classification of stock assets and update the five-level classification information of related debt items; Update the subject rating: modify the data based on the subject rating of the existing assets and update the subject rating information of the relevant debt items; Update write-off information: update the write-off information of related debt items according to the write-off data; S32: superimpose the updated existing business data with the incremental business data to generate full business data for the forecast date.

4. The method for predicting expected credit losses at future points in time according to claim 3, characterized in that: Step S4 includes: S41, based on the risk classification, overdue days, and write-off information of the forecast date, according to the stage division rules in the impairment plan, recalculate the customer-level risk classification and overdue days to determine the impairment stage of the forecast date; S42. According to the impairment stage of the forecast date, different impairment parameters are called to calculate a preliminary impairment result; S43. According to the weights of optimism, benchmark and pessimism, the preliminary impairment results are weighted and summed to calculate the final impairment result of the forecast date.

5. The method for predicting expected credit losses at future points in time according to claim 4, characterized in that: The impairment stages of the forecast date include stage one, stage two and stage three, and step S42 includes: If the impairment stage of the forecast date is stage one, the preliminary impairment result is the expected credit loss for the next 12 months, ECL = PD × LGD × EAD, where ECL is the preliminary impairment result, PD is the forward-looking adjusted one-year probability of default, LGD is the default loss of the corresponding guarantee method, and EAD is the default risk exposure at the current measurement point; If the impairment stage at the forecast date is stage 2, the preliminary impairment result is the expected credit loss during the duration; if the current exposure method is used, ECL = EAD × CPD i ×LGD, where i is the duration, CPD i is the cumulative default probability after forward-looking adjustment in the i-th period; if the exposure method is used in each period, ECL = Σ N i＝1 (EAD i ×MPD i ×LGD i ) / (1+Rate) (i-1) , where EAD i is the default risk exposure of the i-th period, MPD i is the marginal default probability of the i-th period, LGD i is the default loss of the i-th period, Rate is the loan interest rate, i and N are positive integers; If the impairment stage on the forecast date is stage three, the preliminary impairment result is the expected credit loss during the life of the loan, ECL = EAD × LGD.

6. The method for predicting expected credit losses at future points in time according to claim 1, characterized in that: The impairment results of the forecast date and the comparison information with the impairment results of the business data date include: forecast type code, forecast type name, legal entity, asset number, risk exposure / grouping, original five-level classification, original subject rating, original overdue days, original stage judgment results, forecast overdue days, forecast stage results, forecast impairment results, impairment supplementary amount, and impairment change amount.

7. A prediction system for expected credit losses at future points in time, used to implement the prediction method according to any one of claims 1 to 6, characterized in that: The prediction system comprises: The impairment forecast module is used to create a new forecast plan, which includes: legal entity, impairment plan, forecast dimension, business data date and forecast date; A data processing module, used to prepare business data required for prediction, the business data including: stock business data on the business data date, stock asset five-level classification modification data, stock asset subject rating modification data, write-off data and incremental business data; The impairment prediction module is further used to generate the full amount of business data on the forecast date according to the preset rules and the business data; and to call the impairment calculation function according to the full amount of business data on the forecast date, the impairment plan and the forecast date to calculate the impairment result on the forecast date; The statistical query module is used to query the impairment result of the forecast date and its comparison information with the impairment result of the business data date.

8. The prediction system for expected credit losses at future points in time according to claim 7, characterized in that: The stock business data on the business data date include: impairment risk exposure business data of corporate loans, retail loans, corporate off-balance sheet, credit cards, interbank business, bond investments, non-standard investments, and bill business; The five-level classification modification data of stock assets includes: new five-level classification information of the stock business data to be adjusted, including: legal entity, forecast date, five-level classification, and asset number; The existing asset subject rating modification data includes: new subject rating information of the existing business data to be adjusted, including: legal entity, forecast date, subject rating, asset number, name of transaction subject / issuer / financier / accepting bank; The write-off data is data information of the assets to be written off, including: legal entity, write-off date, and asset number; The incremental business data refers to the new business data added on the forecast date.

9. The system for predicting expected credit losses at future points in time according to claim 8, characterized in that: The impairment prediction module is specifically used for: Updating the stock business data in the business data according to the preset rules includes: Update overdue days: Calculate the time interval between the business data date and the forecast date, assuming it is n days. If the current debt overdue days is 0, then the debt overdue days on the forecast date will still be 0. If the current debt overdue days is greater than 0, then the debt overdue days on the forecast date will be the original overdue days + n. Update risk classification: modify data according to the five-level classification of stock assets and update the five-level classification information of related debt items; Update the subject rating: modify the data based on the subject rating of the existing assets and update the subject rating information of the relevant debt items; Update write-off information: update the write-off information of related debt items according to the write-off data; The updated existing business data is superimposed on the incremental business data to generate the full business data for the forecast date; The impairment prediction module is also specifically used for: Based on the risk classification, overdue days, and write-off information on the forecast date, according to the stage division rules in the impairment plan, recalculate the customer-level risk classification and overdue days to determine the impairment stage on the forecast date; According to the impairment stage of the forecast date, different impairment parameters are called to calculate the preliminary impairment results; The preliminary impairment results are weighted and summed according to the optimistic, baseline and pessimistic weights to calculate the final impairment result on the forecast date.

10. The prediction system for expected credit losses at future points in time according to claim 7, characterized in that: The system further comprises: The impairment parameter module is used to set the rules of impairment parameters and perform corresponding calculations. The calculation results are directly configured and called by the impairment scheme link. The impairment parameters include the following parameters: risk grouping, stage division, PD migration matrix, PD main scale, LGD management, prospective management, EAD management and CCF management. Each parameter sets the corresponding rules and performs measurement; The impairment scheme management module is used to select the risk exposure involved in the impairment scheme. It supports adding risk groups under risk exposure, and then flexibly configures the impairment parameters used under the risk exposure / group and sets the priority. Under normal circumstances, a risk group can use the impairment parameters of the risk group or the impairment parameters of the risk exposure to which it belongs, and cannot use the impairment parameters of other risk groups under the exposure to which it belongs.

Images