CN107657529A

CN107657529A - IFRS9 credit card is expected depreciation method

Info

Publication number: CN107657529A
Application number: CN201711022595.0A
Authority: CN
Inventors: 陶冶
Original assignee: Individual
Current assignee: Individual
Priority date: 2017-10-26
Filing date: 2017-10-26
Publication date: 2018-02-02

Abstract

The credit card of IFRS9 a kind of is expected depreciation method, including financial institution's internal data to being stored with cardholder data pre-processes；Intelligence computation terminal obtains associated external economic data；Intelligence computation terminal synthesis internal data, external economy data carry out depreciation parameter prediction to credit card assets, and depreciation amount is measured and exported.The method that the present invention provides a set of reliable dynamic prediction credit card depreciation amount for the qualified financial institution that holds, accomplish safely and effectively run working lining by intelligent terminal and calculate credit card depreciation number, ensure that management level are implemented monitoring credit card depreciation and changed simultaneously, it is truly realized and timely action is taken to possible credit risk, the air control department for not being only row side provides reliable assessing credit risks, also provides a kind of dynamic depreciation amount based on macroeconomic data forward prediction for Hang Fang Finance Departments.

Description

IFRS9 credit card is expected depreciation method

Technical field

The present invention relates to expected depreciation to calculate, and particularly one kind meets No. 9 texts of IFRS (referred to as IFRS 9) credit card is expected depreciation method, and this method considers internal data and outer for holder's refund wish and loan repayment capacity Portion's macroeconomic data is predicted, and then the loss that future may occur is estimated, suitable for hair fastener qualification Financial institution.

Background technology

IFRS council (IASB¹) require to perform new financial instrument criterion IFRS from January, 2018 9²Substitute existing criterion IAS 39³.Compared with the latter, main one of change is measured and supervisor using expection depreciation model Structure needs the depreciation reserves of depreciation.In August, 2016, China Ministry of Finance is also according to the convergent route map of criterion with IASB signatures It is proposed convergent IFRS 9 new principle financial instrument part exposure draft, i.e. CAS 22⁴、23⁵、24⁶Revised draft.

With traditional row Kapp and the U.S. compared with, China starts late lending market, in earlier 2000s just progressively Credit rating system is set up, and row card body system was just invented in the U.S. before 30 years, had formd so far perfect (ring joins TransUnion, and Ai Ke flies Equifax and benefit wins farsighted tri- big references of Experian for credit rating system and credit information service Office).Currently, China's credit system is not perfect, integralities and chronicity of data are still to be strengthened, along with China's Financial market Developing and opening is relatively later, and trackability need further to improve on the paces for follow International Controller, cause China on IFRS 9 is expected the research of depreciation model and application far lags behind western countries.

Although credit cards product from loan ceiling, is not belonging to the main force in current conventional banking facilities lending business, But with continuous improvement of the individual consumer to credit cards product understanding degree, domestic-investment row credit card

Circulation hits new peak repeatly, at the same time, major foreign capitals row successively China obtain independent hair fastener power (star-spangled banner 2012, Slag is made a call to 2014,2016, Huifeng) and actively develop credit card business, along with major internet financing corporation pair in recent years " virtual credit card " market (Jingdone district informal voucher, ant flower, Suning is wilful pay) is constantly beached, and whole credit card purchase colony is just Go from strength to strength, it is also constantly soaring to the occupation rate of credit market.The non-mortgage class credit main body of retail as future society, credit The depreciation of card product, which is assessed, highly to be paid attention to.

In view of the IFRS 9 international effective date is on January 1st, 2018, so far, China so it is international use it is logical It is still a kind of model based on incurred losses with depreciation method, i.e., has objective evidence to show that loan has been sent out in balance sheet During raw depreciation, it can just be provided a loan depreciation according to the amortized cost of loan and the subtractive value of current value of future cash flow.The party Method has obvious parent periodically, and economically the departure date, loan defaults rate and loss late are relatively low, the loan loss provision of depreciation compared with Few, so as to which profit increases, bank further expands the line of credit, economic continuous prosperity；And the departure date, loan defaults and loss under economical Increase, the loan loss provision of depreciation is also more, and this will further deteriorate the financial situation of bank.As can be seen here, occurred The acknowledging time that loss model prepares for loan loss is later, loan the credit cycle early stage depreciation reserves the amount of money compared with Few, this causes loan early stage interest income to be overestimated, and it is economical descending to absorb that loan loss provision can not obtain effectively accumulation Credit loss caused by period.Therefore, it is necessary to new loan loss provision depreciation model is developed, to alleviate loan loss standard The parent of standby gold is periodically.

It is expected that depreciation model is a kind of method for being used for predicting loan Reserve Fund depreciation amount, in loan whole life cycle Interior consideration loan loss, each accounting end of term accounting subject need to reappraise the expected cash flow of loan and expected credit loss, in advance Any change of phase credit can be reflected in cash flow present worth, and influences profit and loss change.The model can identify earlier With confirmation credit loss, contribute to the accumulation that loan loss prepares to tackle a large amount of loan losses that economic descending period occurs. Solves the close periodic problem in incurred losses model to a certain extent.

At present, the research for being expected depreciation model for 9 times credits card of IFRS both at home and abroad is very few.In the world, only Australia is national The formal exploitation and implementation for externally announcing to have completed the model of one, bank, other banks actively prepare related work； The country, some only researchs are also stayed in the simple hypothesis of conventional interior grading method or roll modeling method, to solving finance Discounted in mechanism promise breaking data deficiencies, the economic downlink data of shortage, the definition of life cycle and switching at runtime, and complete period existing It is still insufficient that golden stream such as estimates at the challenge.

The content of the invention

It is an object of the invention to provide one kind to meet No. 9 texts (hereinafter referred to as IFRS 9) of IFRS Credit card is expected depreciation method, to solve at least one above-mentioned technical problem.

The technical solution of the present invention is as follows：

A kind of credit card is expected depreciation method, comprises the following steps：

Step 1, the financial institution's internal data for being stored with cardholder data is pre-processed, including holder's state, Holder's accrediting amount, service condition and the term of validity that holds etc.；

Step 2, intelligence computation terminal obtain associated external economic data, including unemployment rate, real GDP become Rate, consumer index rate of change, described intelligence computation terminal are in mobile phone, notebook computer, desktop computer, tablet personal computer Any one；

Step 3, intelligence computation terminal integrate inside and outside portion's data and carry out depreciation parameter prediction to credit card assets, and finally Depreciation amount is measured and exported；

The method that the present invention provides a set of reliable dynamic prediction credit card depreciation amount for the qualified financial institution that holds, And accomplish that management level can be implemented to monitor the variation of credit card depreciation by intelligent terminal, it is truly realized and possible credit risk is adopted Take timely action.

Brief description of the drawings

Fig. 1 is the schematic flow sheet that credit card of the present invention is expected depreciation method.

Fig. 2 is that the expected depreciation reserves of the present invention collects schematic flow sheet

Fig. 3 is the structured flowchart that credit card of the present invention is expected depreciation method

Embodiment

In order that the technical means, the inventive features, the objects and the advantages of the present invention are easy to understand, tie below The present invention is further elaborated for conjunction embodiment and accompanying drawing.

One kind meets IFRS 9 and supervises defined credit card expection depreciation method, comprises the following steps：

1) internal data of storage is pre-processed, including：

1.1) data scrubbing, missing values, outlier processing and Data concentrating：

The core data of the credit card such as He of form 1

Shown in form 2, including：Account-related information (account number, account status, pause code), account validity letter Cease (day of opening an account, credit card Expiration Date), the account accrediting amount, account cost information (account retail remaining sum, account closing balance, Credit card loan balance), overdue information (0-29,30-59,60-89,90-119,120-149,150-179,180-209, 210 days and the overdue amount of money of the above), interest rate information (account refund interest rate), the actual depreciation information of history (exceedes for 90,120 and 150 days Phase), check and write off and reclaim total value.

Existence account is retained by the screening to account status and pause code, according to model requirements, by opening an account Day and the limitation of Expiration Date, obtaining some specific analysis month YYYYMM, (ratio then needs if desired for analysis 1 month information in 2017 Obtain 201701 related datas) credit card existence account assets information.The credit card essential information data of form 1. are detailed and internal Processing 1 (Default Probability is open related to default risk)

The credit card essential information data detail 2 of form 2. (loss given default is related)

Above-mentioned core data by missing values, exceptional value processing after, extract feature for the foundation of expected depreciation model and become Amount, is handled as shown in Table 3：

Defining qualified relevant three conditions for surviving account features variable IS_VALID is respectively：

(1) ACCOUNT_STATUS is 0 or 1 or 2 (i.e. non-cancellation accounts),

(2) ACCOUNT_BLOCK_CODE is not equal to B (i.e. non-freezing account), and

(3) X90_PLUS_CURRENT_BALANCE is not more than zero (i.e. non-90 days or more delinquent accounts), if the above three Individual condition all meets that it is 1 then to set IS_VALID values, is otherwise 0；

If (i.e. non-90 days or more delinquent accounts) X90_PLUS_CURRENT_BALANCE is not more than zero, definition is disobeyed About label IS_DEFAULT values are 0, and label value of otherwise breaking a contract is 1；

If 90_PLUS_CURRENT_BALANCE is more than zero, phase III identification characteristics variable IS_STAGE_3 is defined It is otherwise 0 for 1；

If X30_89_CURRENT_BALANCE is more than zero, defining second stage identification characteristics variable IS_STAGE_2 is 1, it is otherwise 0；

If PAST_DUE_AMOUNT is more than zero, it is 1 to define first stage identification characteristics variable IS_STAGE_1, otherwise For 0；

Define credit card utilization rate characteristic variable UTILISATIOIN=ACCOUNT_CURRENT_BALANCE/ ACCOUNT_CREDIT_LIMIT.

The credit card essential information feature extraction of form 3.

1.2) calculating of historical risk parameter：

In history observation rate of violation (Observed Default Rate, the referred to as ODR) definition at M0 moment in some month It is as follows：

ODR_M0=(IS_DEFAULT_M1+IS_DEFAULT_M2+ ...+IS_DEFAULT_M12)/IS_VALI D_M0

Define history loss given default (Loss Given Default Rate, referred to as LGD) on certain month M0 such as Under：

LGD_M0=(CC_IIA_90DPD+CC_IIA_120DPD+CC_IIA_150DPD+CC_WO_AMOUNT-CC _ RECOVERY_AMOUNT)/(OVERDUE_BAL_90DPD+OVERDUE_BAL_120DPD+OVERDUE_BAL_150DPD+CC_ WO_AMOUNT),

Credit card remaining sum ACCOUNT_CURRENT_ (is based on by the weighted average to each history month credit card BALANCE), it can be deduced that the history Default Probability LGD. of credit card total assets aspect

Holder is defined as to the utilization rate of credit card：

UTILISATION=ACCOUNT_CURRENT_BALANCE/ACCOUNT_CREDIT_LIMIT, to above-mentioned use Rate carry out rationally split after, the holder colony on each subregion is further analyzed to obtain on each subregion it is additional because Sub- ADD-ON is 20% (when UTILISATION is 0 to 80%)；10% (when UTILISATION is between 80 to 100%)； 5% (when UTILISATION is more than 100%).So as to show that default risk is open and (Exposure at Default, be referred to as EAD)：

EAD=ACCOUNT_CURRENT_BALANCE+ADD_ON*ACCOUNT_CREDIT_LIMIT.

Discount rate (Discounting Factor, referred to as DF) is calculated by holder's refund interest rate r：DF_N= (1+r) ^N, wherein N are held the life cycle time limit of credit card by holder.In example, refund interest rate r=30% is defined, and The life cycle of credit card is no more than 5 years, i.e. N≤5.

2) importing and processing of external data：

2.1) by historical data analysis, credit card asset portfolio Default Probability and external economy data phase in industry Closing property is higher following three kinds：

The external economy data characteristics of form 4. is extracted

A=Ln [P (1-P)]=- 5.9185+0.1562*UNEMPLOYMENT_RATE+ (- 0.0276) * REAL_GDP_ GROWTH+(-0.1411)*CONSUMER_PRICE_CHANGE

Wherein, UNEMPLOYMENT_RATE represents that total unemployment accounts for the ratio (unit %) of workforce, REAL_ GDP_GROWTH is the variation ratio (unit %) compared with last year actual GDP, CONSUMER_PRICE_CHANGE be with it is upper The ratio that year consumer index is compared changes (unit %).

And then P value is calculated, P=exp (a)/[1+exp (a)], P value is bigger, i.e., outside Default Probability is got over It is high.

2.2) following new variables is exported, is calculated available for follow-up：

The new variables brief summary of form 5.

The actual Default Probability of prediction future inter (PD_M, PD_M_S2, PD_M+1_S2, PD_M+2_S2, PD_M+3_S2, PD_M+4_S2), it is described as follows：

The term Anticipatory breach probability of first stage holder (the overdue holder for being no more than 30 days) is：

PD_M={ P_M/P_M-1 } * ODR_M-1,

Wherein, outside Default Probability when P_M and P_M-1 is illustrated respectively in month M and M-1 (has specific Jie in above-mentioned 2.1 Continue P_M P i.e. on month M computational methods), ODR_M-1 represents the actual rate of violation in inside in month M-1.

The lifetime Anticipatory breach probability P D_M_S2 of second stage holder (the overdue holder between 30-89 days)：

It is derived from by following two equation：

PD_M={ P_M/P_M-1 } * ODR_M-1, and PD_M=PD_M-1+ (1-PD_M-1) * PD_M_S2, i.e.,：

PD_M_S2={ [P_M/P_M-1] * ODR_M-1- [P_M-1/P_M-2] * ODR_M-2 }/(1-PD_M-1).

3) collect expected depreciation reserves and export

It is expected that depreciation reserves classified calculating method is as shown in the table：

Form 6. is expected depreciation reserves classified calculating method

First stage is expected depreciation amount：ECL_S1=PD_M*LGD*EAD

Second stage is expected depreciation amount：ECL_S2=PD_M_S2*LGD*EAD*DF_1+PD_M+1_S2*LGD*EAD* DF_2+…+PD_M+4_S2*LGD*EAD*DF_5,

Phase III is expected depreciation amount：ECL_S3, the Reserve Fund under former IAS39 methods is directly continued to use,

By the classified calculating of above-mentioned account aspect, the depreciation reserves that any account should be assigned to is obtained, through expection Depreciation reserves collects flow, the depreciation reserves total value of credit card asset portfolio aspect is obtained, referring to Fig. 2.

Empirical tests, certain foreign capitals row credit card target magnificent that such scheme is directed under economic good situations carry out depreciation meter Calculate, the depreciation under the frameworks of IFRS 9 is compared with real credit card loss of assets, and it is about 40% to guard degree, and existing IAS 39 Then reach 70%, consider from fund cost angle is saved, the depreciation risk identification Du Genggao under new departure, conservative can also obtain To guarantee.

Intelligence computation terminal is provided with a wireless communication module, and intelligence computation terminal is stored with by wireless communication module connection The database of credit card user asset portfolio data, including holder's user profile after encryption, account validity information, account The accrediting amount, account cost situation and overdue and check and write off information, meanwhile, by wireless communication module, intelligent terminal can also call External economy data, including the total unemployment of proce's-verbal account for the ratio of workforce, the variation ratio compared with last year actual GDP Rate and the ratio compared with last year consumer index change.

After all data input intelligence computation terminals, after described intelligence computation terminal is by data processing and assembling, Be deposited in caching in etc. it is to be called, because internal data has been encrypted, ensure that the security of data acquisition.

Described intelligence computation terminal is provided with prediction depreciation computing module, and the module is transferred provided with data, data calculate, be individual Six example analysis, result displaying output, parameter setting and LOG daily records sub-modules, wherein：

Data are transferred：Button, interface expansion display are transferred by click data：

(1.1) data are imported, and inactive data in the buffer is had been introduced into before can now being imported by the button；

(1.2) null value inspection, by some rules set in advance, Basic examination is carried out to importing data, it is necessary to pass through Basic examination could carry out after flow；

Data calculate：After data are transferred successfully, by data computed push-buttom, bottom is carried out by scheme described above (form 6) Layer calculates.

It is worth noting that, intelligence computation terminal is by the relation between the Algorithm Analysis holder of data mining and promise breaking, The feature being had a significant impact to promise breaking is selected, and analyzes refund wish of the external economy environment on the colony that holds to influence, to holding People's default risk carries out dynamic evaluation.By the algorithm of data mining, what is optimized fits, and draws inside and outside relation, makes The matching degree highest of assessed value and actual promise breaking record.The method that prediction uses regression analysis and P value hypothesis testings.Individual example point Analysis；After the completion of calculating, intelligence computation terminal is optional to enter an example assay surface, and before credit card risk exposure ten client is carried out The displaying of each intermediate variable and last result of calculation, one is provided directly for row side user of service (especially air control or financial employee) The depreciation demonstration of the client that has a great influence seen.

As a result displaying output：Concentrate and show total depreciation and the classification depreciation number on first, second and third stage.Simultaneously can be with Selection generation report, the result of account aspect to asset portfolio aspect is exported to file (.csv .txt or .xlsx form) and joined Number is set：A solution is provided for some advanced users of services' (needing to carry out variation renewal to parameter) of intelligence computation terminal Scheme；LOG daily records：For recording main step result, if wrong, user can go LOG daily records to leaf through, convenient positioning The step that malfunctions or reason.

It should be noted that the intelligence computation terminal is provided with an identification system, for user class management (readable, writeable, risk parameter can be changed), described identification system include camera with pronunciation receiver at least A kind of signal acquisition module；

Described identification system gathers at least one of face, iris, business card information by the camera；

The identification system gathers sound by the pronunciation receiver；

The identification system also includes a microprocessor system, and the signal acquisition module connects the miniature place Device system is managed, the microprocessor system is provided with a signal output part；

When microprocessor system judges that the personal information that signal acquisition module collects has prediction depreciation measuring software When logining authority, described microprocessor system control depreciation measuring software automatic sign in；

The device number for logining prediction depreciation software for calculation is sent to server group by described microprocessor system, is realized The intelligence computation terminal for logining signal scoring software transfers the data of server group.

The general principle and principal character and advantages of the present invention of the present invention has been shown and described above.The technology of the industry Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the simply explanation described in above-described embodiment and specification is originally The principle of invention, without departing from the spirit and scope of the present invention, various changes and modifications of the present invention are possible, these changes Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its Equivalent thereof.

Claims

1. the credit card of IFRS9 a kind of is expected depreciation method, equipment includes intelligence computation terminal, wireless communication module and is stored with The database of credit card user asset portfolio data, described intelligence computation terminal by described wireless communication module with it is described Database be connected, described intelligence computation terminal is provided with identification system, and the identification system includes signal acquisition mould Block, microprocessor system, the described microprocessor system of described signal acquisition module connection, the microprocessor system System has signal output part and display screen, and described microprocessor system has prediction depreciation software for calculation；Characterized in that, This method comprises the following steps：

1) internal data of the cardholder information of the intelligence computation terminal-pair storage described in is pre-processed：The intelligence computation is whole End is stored with the database of credit card user asset portfolio data by wireless communication module connection, after calling includes encryption Holder's user profile, including the account number of encryption, status of credit card, credit card pause code, day of opening an account, the Expiration Date, Credit volume, account retail remaining sum, account closing balance, account loan balance, overdue information are (without overdue, 0-29 days, 30-89 My god, 90 days and the above it is overdue), refund year interest rate, history depreciation information (90,120 and 150 days overdue), check and write off and reclaim always Volume, described pretreatment include holder's state, holder's accrediting amount, service condition, the cleaning for the term of validity that holds, missing, Outlier processing, Data concentrating, the refinement of characteristic and the processing of historical risk parameter；

2) the intelligence computation terminal described in actively transfers associated external economic data by described wireless communication module, including loses Industry rate, GDP rate of change, consumer index rate of change,

All data after intelligence computation terminal is accessed, by data processing and assemble after, be deposited in caching in etc. it is to be called；

3) internal data described in intelligence computation terminal synthesis, external economy data described in carry out depreciation to the assets of credit card Parameter prediction, depreciation amount is measured and exported by the output interface of described intelligence computation terminal：It is total including depreciation Volume, first stage, second stage, the depreciation number on the phase III, the form of account aspect to asset portfolio aspect, including .csv .txt or .xlsx forms.

2. IFRS9 according to claim 1 credit card is expected depreciation method, it is characterised in that described intelligence computation is whole End is mobile phone, notebook computer, desktop computer or tablet personal computer.

3. IFRS9 according to claim 1 credit card is expected depreciation method, it is characterised in that described signal acquisition Module is camera or pronunciation receiver.

4. IFRS9 according to claim 1 credit card is expected depreciation method, it is characterised in that described internal history letter Breath, including history actual observation rate of violation, history loss given default, account utilization rate, export based on history default risk is open Each subregion on additional factor, average discount rate and life cycle duration.

5. IFRS9 according to claim 1 credit card is expected depreciation method, it is characterised in that described data mining Analysis includes：

1) formula between outside macroeconomy information and industry Default Probability is：

A=Ln [P (1-P)]=- 5.9185+0.1562*UNEMPLOYMENT_RATE+ (- 0.0276) * REAL_GDP_GROWTH +(-0.1411)*CONSUMER_PRICE_CHANGE；

Wherein, UNEMPLOYMENT_RATE is that the total unemployment of proce's-verbal accounts for the ratio (unit %) of workforce, REAL_GDP_GROWTH is variation ratio (unit %) compared with last year actual GDP, CONSUMER_PRICE_CHANGE (unit %) is changed for the ratio compared with last year consumer index；And then P value is calculated by following equation：

P=exp (a)/[1+exp (a)], P value is bigger, and outside Default Probability is higher；

2) prediction the actual Default Probability of future inter (PD_M, PD_M_S2, PD_M+1_S2, PD_M+2_S2, PD_M+3_S2, PD_M+4_S2 it is) as follows：

PD_M={ P_M/P_M-1 } * ODR_M-1,

Wherein, P_M and P_M-1, which is illustrated respectively in, (has specific introduction in M months and the outside Default Probability in M-1 months in above-mentioned 2.1 P_M is the computational methods of P on month M), ODR_M-1 represents the actual rate of violation in inside in month M-1.

The lifetime Anticipatory breach probability of second stage holder (the overdue holder between 30-89 days) is：

PD_M_S2=={ [P_M/P_M-1] * ODR_M-1- [P_M-1/P_M-2] * ODR_M-2 }/(1-PD_M-1).

6. IFRS9 according to claim 1 credit card is expected depreciation method, it is characterised in that described intelligence computation The depreciation measuring software of terminal is transferred provided with data, data calculate, individual example is analyzed, result shows output, parameter setting and LOG days Will six functionses.