Home

User Manual: Product Name

Contents

1. Income Economic Capital Allocated This is the profitability percentage comparable to minimum risk adjusted benchmarks or hurdle rates SVA The Shareholder Value Added SVA also sometimes referred to as Economic Value Added EVA is defined as Economic Net Income Economic Capital Allocated Hurdle Rate This is the value which combines both percentage profitability and size of transactions to find whether a transaction or a sub portfolio creates or destroys value for shareholders RBP Risk Based Price RBP is defined as the product of the capital required for a new portfolio member and the hurdle rate Oracle Financial Software Services Confidential Restricted 46 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 ORACLE Oracle Financial Services Economic Capital Advanced 1 1 2 0 0 User Guide May 2014 Version number 1 0 Oracle Corporation World Headquarters 500 Oracle Parkway Redwood Shores CA 94065 U S A Worldwide Inquiries Phone 1 650 506 7000 Fax 1 650 506 7200 www oracle com financial_services Copyright 2014 Oracle and or its affiliates All rights reserved No part of this work may be reproduced stored in a retrieval system adopted or transmitted in any form or by any means electronic mechanical photographic graphic optic recording or otherwise translated in any language or computer language without the prior written permission of Oracle Financi
2. crosta ey a TAPA 7013 03 0424 pu Wa Last Wosited By Figure 13 Cash Flow Engine Cash Flow Engine Bucket Definition Cash Few Bqugtons gt Caah Pow togra Bucket Oetntes tot Mode Exposure Equation Detinitica Exposure Equation Name Selectos Dalovat_ Orts Set Exposure O n Bucket Properties Debervemect Date anote 200 TE Total Number of Oucrets 1 Time Buckots dle Tine Ductets Yj 29 2AC 3010 To ooze _ Kqwaten 21 060 2010 To 20 66 2019 de Eauaten Hare Equation Type foston Y Ottow Lausten Data 0 12 Qrosa Doneste Product Growth Rate 25 hien Egunen Piw D 12Oras Doneste Product Growth Ratej 15 User ita Crested By vs IP APR203 030524 PU G mA Lant uostas By Figure 14 Cash Flow Engine Oracle Financial Software Services Confidential Restricted 18 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Cash Flow Engine Equation Figure 15 Cash Flow Input Screen After all the data entries are updated you need to define a Cash Flow Model in the Model Definition screen like the above defined techniques You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation The input or the output variables are defined by selecting the relevant variable in the modeling framework screen variable tab The following figure sh
3. Cash Flow Model Only Specialized Lending All model definitions are done on a set of exposures affected by the same set of variables Dataset should be selected specific to the model defined Data set filter is used to group the exposures into one set e Technique 1 Linear Regression Unlike unconditional PD unconditional LGD has only 3 methods for computation of which Linear Regression with random effects can be used for Corporate Sovereign Bank and Retail Asset Class Retail also uses Historical Default Weighted Average of Pool observed LGD technique You can choose two regression techniques for the calculation of LGD Linear Regression with Maximum Likelihood ML Linear regression with Restricted Maximum Likelihood REML The processes to define these two models are the same The flow to define linear regression model is explained as follows Objective The objective of the model is to calculate the unconditional LGD by Linear Regression with fixed and random effect You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation The data set on which the LGD modeling is to be done should be selected The input or the output variables are defined by selecting the relevant variable in the modeling framework screen Model Name Linear Regression Model Description Linear Regression for calculation of U
4. O Varade Nare Courserpaty Detest s 255495505208 Courtergarty Current assets Cogrterperty Sietum viTeretiroeeT gt Figure 8 Variable Addition Logistic Regression in the modeling framework does not require parameters input like MCEM model Process outputs for the Model are shown below Mage Mann gt oei Det leew Mote 2 Modei Deta a ose tamer osa Descreoten 110 14 ao Parameter Estare Oevance c Qon On Error Degrees of Freedom Error Degrees of Freecom Cussicaton Tatie end ROC Report Clasaificanen Table ang ROC Reson Vodei Preitormance Report Model Ner formare Report Cut Ott Procanity Bases Descrotve CLON Propanary Daes Oracrenre Tes tr Autacorreiaben ROC Curve Curabve Gan Chart Reston PP Pot Process Outeut Requred Tanies 3 By Presetes Pot sepensent Vanabies Dy Rescue Mot t User into Created Dy Figure 9 Process Output Process Output Modeling framework supports the following process outputs for logistic regression Beta Coefficients Classification Table Oracle Financial Software Services Confidential Restricted 14 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Deviance Model Performance Report ROC Report Residuals Standard Error odds ratio p values t statistics You can choose any or all of the above parameters as process outputs The process outputs calculated are the coefficients
5. holding period is generally assumed to be one year Conditional VaR based Economic Capital Credit Risk The expected loss in situations worse than a user chosen quantile of the loss distribution of the customer one holding period later The holding period is generally assumed to be one year Unconditional Probability of Default The probability of customer defaulting one holding period later given the information available today The holding period is generally assumed to be one year Unconditional Loss Given Default The loss anticipated if the customer happens to default one holding period later given the information available today The holding period is generally assumed to be one year Marginal Risk Contribution The change in Portfolio Economic Capital if a portfolio member is added to it removed from it This is equivalently also referred to as Incremental Economic Capital requirement Absolute Risk Contribution Change in Portfolio Economic Capital given a small change in the size of a portfolio member This is equivalently also referred to as Component Economic Capital requirement Regulatory Capital Credit Risk Capital required for a portfolio member as per the Basel II norms Economic Net Income Revenues from the Performance System less the Allocated Economic Capital Hurdle Rate The minimum benchmark return expected for a portfolio member RAROC The Risk Adjusted Return on Capital RAROC is defined as Economic Net
6. is required to be updated The target column is in the FACT table and has a relationship with the table on which the hierarchy is based The target hierarchy must not be based on the FACT table Mapping This is an operation classifies the final record set of the target that is to be updated into multiple sections It also encapsulates the update logic for each section The logic for the update can vary depending on the hierarchy member business processor used The logic is defined through the selection of members from an intersection of a combination of source members with target members Node Identifier This is a property of a hierarchy member In a Rule definition the members of a hierarchy that cannot participate in a mapping operation are target members whose node identifiers identify them to be an Others node Non Leaf node or those defined with a range expression Refer to the section Defining Business Hierarchies in the Unified Metadata Manager Manual for more details on hierarchy properties Source members whose node identifiers identify them to be Non Leaf nodes can also not participate in the mapping Building Process Blocks for treatments Introduction to Processes A set of rules collectively forms a Process A process definition is represented as a Process Tree The Process option in the Rules Framework Designer provides a framework that facilitates the definition and maintenance of a process A hi
7. model Counterparty Equity weekly returns standard deviation current market price of Equity of the above counterparty and Debt amount of the counterparty are selected as Input variables In the variable name tab you have to browse the measures for defining the input variables Refer to the following screenshot Model Definition Coderparty PO Estmaten Corporate We Tha model a used w estimate Probability of Detaut for Corporate Econom Captal Vera usse Non See Exposures tor Medeing Sampling meuts Data Outputs Variable Selection El vaca Gee Counterparty Equty weekly returra standard deviates Dett ameur of me Courterpeny Destasce To Detast Current Wartet Pree of Equey of Otagor Usec info Crested By ITAPRION 11 47 25 AU Figure 3 Variable Definition in Models To run Merton model you will have to specify two parameters Parameters Oracle Financial Software Services Confidential Restricted 8 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 e Weekly risk free rate at which one can borrow or lend in the market r Time Time 52 Weeks 365 7 considering there are 7 trading days Process Outputs The technique gives the following output Distance to Default DD DD V F o The computed DD is mapped to the closest value in the mapping table of DD and Unconditional Probability of Default UC PD A tabulated one to one mapping between variou
8. of the CDM model is VaR CVaR and Expected You need to select the measure corresponding to VaR CVaR EL and UL from the list in the variable name tab Loss Model 3 Time to Default Model This model is used for the OTC SFT asset class to calculate the loss As a pre requisite for this loss model the unconditional model for PD exposure pool and LGD should have been defined and deployed Objective The objective of CDM is to calculate the risk factors like Conditional PD Conditional LGD EAD VaR EL and UL The prediction for Conditional PD and Conditional LGD You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation Refer the screenshot below Technique selected is Time to Default Model and dataset selected is Non_Sec_Time_to_Default The dataset filter excludes all but SFT OTC product types for the calculation The input or the output variables are defined by selecting the relevant variable in the modeling framework screen Model Name Time_to_Default Model Model Description Time_to_Default Model for OTC SFT product types to calculate Loss Model Objective EC_Time to Default The Modeling framework screens are as follows Model Definition Figure 26 Time to Default Model Input Oracle Financial Software Services Confidential Restricted 33 User Guide Oracle Financial Services E
9. server with the Model ID To execute the batch select the relevant model id to run the relevant model The screenshot for the same is given below Oracle Financial Software Services Confidential Restricted 34 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 3 Modeing Wedoas here Bor ORACLE Financial Services Analytical Applications Infrastructure mans Request for Mode Execution 1 Mosel z Request for Model Execution Moce Outputs 2 Champion Challenger Figure 27 Model Execution Screen 3 2 12 Model Deployment All the models to be used for the prediction or EC calculation should be deployed from the sandbox infodom to the production infodom The prediction model use these deployed models as their underlying models to calculate EC The screenshots to deploy models are as follows Oracle Financial Software Services Confidential Restricted 35 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Request tor Mode Execution 8 Request for Modei Execution Modei Outputs Champion Challenger Request for Model Execution Moga Execution gt Sequest tor os Hiequesl for Medel Execetion Regner Boren Figure 28 Model Deployment Screen To deploy a model click the relevant model and Request for Deployment and Authorize amp Deploy and save as shown above 3 2 13 Run Definition The Baseline Runs defined in the Oracle
10. training sample is called so because it trains the model and the test sample is used to test the model The framework currently supports random samples or stratified random samples to be taken with or without replacement Specification of training sample is mandatory Sample specifications allows for either absolute value specification or a percentage specification Process Output Process outputs are the set of outputs at a model level You can choose the links for csv files in the Model Output screen to view these Process Outputs These process output provides you with a summary of the model analysis and hence can be used to compare different versions of the same model For Example R for regression Coefficients of regression and so on Data Output Data Outputs are thrown at the record level and stored in the Rev_Model_Output_Details Examples for the data outputs are Predicted PD values from a regression based PD model K Means cluster ID for each record given by the K Means clustering model when we execute a deployed model in the production infodom 3 2 16 Things to Remember e Run Definition You can keep on defining as many numbers of models as you want in the Sandbox Finally you should choose the models that you want to use in the EC calculation Once you are ready with those set of models deploy them into the production infodom Then in the production infodom tie these models in a run For doing so select the model name in t
11. D and Unconditional PD is given as an input variable In the variable name tab you have to browse the measures for defining the input variables You also need to input the percentage influence of general market variable in the following screen Oracle Financial Software Services Confidential Restricted 30 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Varetie Time References fter S Variable Selection varate Figure 24 Variable Addition The parameters and the process outputs for single counterparty of Credit Metrics Structural Model are as follows Parameters Idiosyncratic Percentage of Counterparty Process Output This technique gives the following output The coefficients of the equation for example Al A2 An An 1 Using the coefficients the value of Z is calculated Z and the Unconditional PD given as an input are used to calculate Conditional PD Model Outputs Distance to default is stored in the model output screen You have to note down either the model objective or model code to view the model output Technique 2 Monte Carlo Expectation Maximization for Conditional PD Monte Carlo Expectation Maximization as explained in Technique 3 MCEM to calculate UCPD under the heading Calculation of Unconditional PD is used to calculate the coefficients of regression model This technique is selected only when the unconditional PD is calc
12. Financial Services Economic Capital Release 1 1 2 0 0 Product are given in the run chart document accompanying the installer Each Run has a chain of processes that may consist of Models DTs Computation and Insertion rules You can define relevant processes and run conditions while defining the Run to match the requirement The Business Model Execution Conditional Default Model Time to Default Model and so on and underlying models execution are initiated through the Run 3 2 14 Run Execution After the Run is defined in the production infodom click the Manage Run tab to Request for Execution of the Run In the New Request for Execution page input the relevant MIS DATE and the Status of Request of the Run By clicking the Save button a batch is registered which is executed in the Batch Group Execution Screen 3 2 15 Important Notes Sampling Oracle Financial Services Advanced Analytical Framework also supports sampling Sampling is an essential and critical part in modeling It allows for three types of samples to be taken Oracle Financial Software Services Confidential Restricted 36 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Training Sample Control Sample Test Sample A training sample is the sample on which the model is configured and a test sample is the sample on which the model gets executed predicted Control sample is a sample which can be used for model comparison Therefore
13. M is the prediction model where the conditional PD and LGD are predicted by the use of simulated values of variables Conditional Default Model CDM also computes the Value at Risk Conditional Default Model PD LGD EAD Approach is used for the Corporate Sovereign Bank and Retail Asset Class Specialized Lending Securitization and OTC SFTs use different prediction models Objective The objective of CDM is to calculate the risk factors like Conditional PD Conditional LGD EAD VaR EL and UL The Modeling framework screens are as follows Oracle Financial Software Services Confidential Restricted 27 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Figure 21 Conditional Default Model Definition Screen You need to define the Model Name Model Description and Model Objective in the modeling framework screen The screen asks for the technique and a relevant data set to be used in the calculation The data set on which conditional modeling is supposed to be done is selected from the drop down list The screen has two tabs Input and Output Variable Model Name Conditional Default Model CDM Model Description CDM for PD LGD EAD and VaR calculation Model Objective Loss Calculation Input Inputs for conditional default model are Simulation Model select the simulation multivariate model ID used for predicting the general market variables PD Models input th
14. Oracle Financial Services Economic Capital Advanced User Guide Release 1 1 2 0 0 May 2014 ORACLE FINANCIAL SERVICES User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Contents ABOUT THE GULDE iis ccs caccdedacetscolededadelacsdedeSevecedsGocededeSededcceUecodddelecvelasevevescvesedeGedevedasedevedesecovevesess III SEOPE OR THE GUIDE AE E vaeth os tas cab tet duwseaeeuedl ae echoes tune tens sant oageaude an eeeueate ota tens N ll AUDIENCE ia dd a O TEES AS Ill DOCUMENT CONVENTIONS iia o Ill I INTRODUCTION A os 1 1 1 ANALYTICAL APPLICATIONS OVERVIEW s0sscscscccsccscsssscscessecuccevsssscesscescssseesscssecncessessscssecncceacessesscesseescessssscessecnees 1 2 INSTALLING THE ORACLE FINANCIAL SERVICES ECONOMIC CAPITAL ADVANCED e E aa ug eR Ea EN Dn Bs bak Soh ee 2 2 1 M DEL UPLOAD a dai 2 3 ORACLE FINANCIAL SERVICES ECONOMIC CAPITAL ADVANCED cccsssccesseeees 3 3 1 SCOPE OF ORACLE FINANCIAL SERVICES ECONOMIC CAPITAL ADVANCED cooooccccncnnnnnnooccnnnnnnnnnnonornnnnnnncnnnnnrrncrnnnnnnnnnnoss 3 3 2 ECONOMIC CAPITAL PROCESS ELO Wes acia 3 3 2 Lo VSGNADOX DEFINITION EE E a EEE E 4 3 2 2 Variable Prepara O siciliana den aa de aauaa ada ra aa aa E a sess 4 3 2 3 Model D finitioN senenn enaa ode a oa ea aa iaa dE ca ead Redan AEE 4 3 2 4 Cale lation of Unconditiona PD seccccececcssceszecssvecitecdenvesttaacdeceticedeutss a a aaie iaa ii aiai iiti 6 3 2 5 Calculation of Unconditio
15. Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition ree fey Versate TmeRaterenceg fier Sampling hots ProcessO Figure 18 Process Output for LGD Model Outputs All the parameters are stored in model output screen You have to note down either the model objective or model code to view the model output Technique 2 Historical Default Weighted Average of Pool Observed LGD This technique is defined only for Retail Asset Class to calculate the unconditional LGD In a retail pool PD LGD and EAD of each exposure are provided as a download for different historical dates The LGD of the pool is calculated as the weighted average of LGD of exposure to the defaulted exposures in a pool Pool LGD Y PD no of exposures LGD Y PD no of exposures Technique 3 Cash Flow Model Objective Cash flow model is only used for specialized lending This technique is used to calculate LGD for specialized lending exposures The process to define cash flow model is provided in the section Technique 6 cash flow model to calculate UCPD under the heading Calculation of Unconditional PD Unconditional LGD is calculated from EAD and Loss percentages These calculations are done in the Run Rule framework of the modeling framework 3 2 6 Calculation of EAD EAD modeling in the Oracle Financial Services Economic Capital Advanced Solution all
16. ains one or more fact tables the points rays contain the dimension tables Dimension Table Dimension Table Products Time Geography Sales Fact Table Customer Channel Figure 29 Data Warehouse Schema Fact Table In a star schema only one join is required to establish the relationship between the fact table and any one of the dimension tables which optimizes queries as all the information about each level is stored in a row The set of records resulting from this star join is known as a dataset Metadata A term used to denote data about data Business metadata objects are available to the user in the form of Measures Business Processors Hierarchies Dimensions Datasets and Cubes etc The commonly used metadata definitions in this manual are Hierarchies Measures amp Business Processors Hierarchy A tree structure across which data is reported is known as a hierarchy The members that form the hierarchy are attributes of an entity Thus a hierarchy is necessarily based upon one or many columns of a table Hierarchies may be based on either the fact table or dimensional tables Measure A simple measure represents a quantum of data and is based on a specific attribute column of an entity table The measure by itself is an aggregation performed on the specific column such as summation count or a distinct count Business Processor This is a metric resulting f
17. al Services Software Limited Due care has been taken to make this Oracle Financial Services Economic Capital Advanced User Guide and accompanying software package as accurate as possible However Oracle Financial Services Software Limited makes no representation or warranties with respect to the contents hereof and shall not be responsible for any loss or damage caused to the user by the direct or indirect use of this User Manual and the accompanying Software System Furthermore Oracle Financial Services Software Limited reserves the right to alter modify or otherwise change in any manner the content hereof without obligation of Oracle Financial Services Software Limited to notify any person of such revision or changes All company and product names are trademarks of the respective companies with which they are associated Oracle Financial Software Services Confidential Restricted 47
18. ation Historical Scenario Based Pool Default Rate Retail Only 3 2 9 Calculation of Conditional LGD Conditional LGD can be derived from any of the two approaches described below Linear Regression Historical Scenarios Based Pool Retail Only The above techniques are explained in the Loss Models because in Economic Capital product the Conditional PD and Conditional LGD predictions are done in the Loss Models Oracle Financial Software Services Confidential Restricted 26 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 If you choose MCEM for the Unconditional PD calculation then choose the same model MCEM for Conditional PD as well If you choose Merton Model or Transition Matrix Model as the Unconditional PD model then you will have to necessarily choose the Credit Metrics Structural Models CMSM as the Conditional PD model Similarly if you choose Historical Average Default Model retail asset class as the Unconditional PD model then you have to mandatorily choose the same as the Conditional PD model 3 2 10 Loss Model Calculation There are three loss models in Economic Capital Product Conditional Default Model PD LGD EAD Approach e Conditional Default Model Distribution Fitting Time to Default Model only OTC SFT The above techniques are explained in the following sections e Loss Model 1 Conditional Default Model PD LGD EAD Approach Conditional Default model CD
19. cle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Fae bouts Veste Dala Outputs Model Variable Selection Varsle Figure 12 Cash Flow Modeling To define a new cash flow equation click the new button To edit a current model defined select a model and click the edit button In exposure equation definition section input the Cash Flow Equation name and its description Select the data set on which the model will be applied The exposure ID and the hierarchy are also selected to run the model on a specific set of data Refer to the following screenshot You also need to input the Bucket Properties in the relevant screen You need to enter the properties like Disbursement Date Time Bucket Type and Total Number of Buckets For example If time bucket type is monthly and total buckets is 4 then four thirty day bucket is displayed on the screen By clicking Display Bucket button you can enter the Default Threshold Asset Value and Exposure Amount for each bucket Oracle Financial Software Services Confidential Restricted 17 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Cash Flow Engine Bucket Definition Exposure Equation Detinition Exposure Cquaton Name A Selected Oaleset Dats Set Exposure O Bucket Properties Desburbemect Date Total Number of Buckets Hime Buckets da Tee Decreta EEEE poutine tafa
20. conomic Capital Advanced Release 1 1 2 0 0 Simulation Model selects the simulation model that generates error terms of general market variables The simulation model is a multivariate simulation model PD Models select from the list of PD model to be used for the prediction of conditional PD Measure Nettable Pool Peak Exposure Amount select the measure which stores the peak exposure value by clicking the drop down list Measure Time to Default This measure stores the time to default intermediate value calculated by time bucket upper bound or total time bucket VaR Percentile VaR percentile is provided as an input by you Enter the percentile at which VaR should be calculated LGD Model ID select from the list of LGD model to be used for the prediction of conditional LGD Output The output of the CDM model is VaR CVaR and Expected You need to select the measure corresponding to VaR CVaR EL and UL from the list in the variable name tab These loss calculations are also known as Prediction Models The prediction models are executed in the production infodom The underlying models should be executed and deployed in the production infodom for the prediction model to be run 3 2 11 Model Execution After all the PD LGD EAD models are defined it is requested for execution You can define a single model for a technique and execute it While you Request for Execution a batch is registered in the ICC
21. d on datasets that contain a single FACT table The values in one or more columns of the FACT tables within a dataset are transformed with a new value Source This component determines the basis on which a record set within the dataset is classified The classification is driven by a combination of members of one or more hierarchies A hierarchy is based on a specific column of an underlying table in the data warehouse model The table on which the hierarchy is defined must be a part of the dataset selected One or more hierarchies can participate as a source so long as the underlying tables on which they are defined belong to the dataset selected Target This component determines the column in the data warehouse model that will be impacted with an update It also encapsulates the business logic for the update The identification of the business logic can vary depending on the type of rule that is being defined For type 3 rules the business processors determine the target column that is required to be updated Only those business processors must be selected that are based on Oracle Financial Software Services Confidential Restricted 40 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 the same measure of a FACT table present in the selected dataset Further all the business processors used as a target must have the same aggregation mode For type 2 rules the hierarchy determines the target column that
22. der Value Added are also computed for the management lines of business 3 2 Economic Capital Process Flow EC uses modeling techniques available in Oracle Financial Services Advanced Analytics Infrastructure 7 3 3 2 0 The Economic Capital EC calculation and reporting broadly involves the following activities Oracle Financial Software Services Confidential Restricted 3 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Sandbox Definition Variable Preparation Model Definition Model Execution Model Deployment Run Definition Run Execution Report Generation 3 2 1 Sandbox Definition Within production infodom you have to create sandbox infodom Sandbox is defined inside the newly created infodom You have to define the relevant Sandbox Name Sandbox Description Sandbox Location and the relevant dataset for the sandbox The data model is copied to the production once the sandbox is defined The data would be extracted or imported from Production infodom based on the dataset defined there In this step we get data for all raw attributes for a particular time period table A super data set is created with the help of which all the models are defined 3 2 2 Variable Preparation For modeling purposes you need to select the variables required for modeling You can select and treat these variables in the Variable Management screen You can define variables on Measures Hierarchies
23. e End to End processes that are executed Simulation Runs are those scenario End to End processes that are executed Simulation Runs are compared with the Baseline Runs and therefore the Simulation Processes used during the execution of a simulation run are associated with the base process Building Business Processors for Calculation Blocks This section describes what a Business Processor is and explains the process involved in its creation and modification The Business Processor function allows you to generate values that are functions of base measure values Using the metadata abstraction of a business processor power users have the ability to design rule based transformation to the underlying data within the data warehouse store Refer to the section Defining a Rule in the Rules Process and Run Framework Manual for more details on the use of business processors What is a Business Processor A Business Processor encapsulates business logic for assigning a value to a measure as a function of observed values for other measures Let us take an example of risk management in the financial sector that requires calculating the risk weight of an exposure while using the Internal Ratings Based Foundation approach Risk weight is a function of measures such as Probability of Default Loss Given Default and Effective Maturity of the exposure in question The function risk weight can vary depending on the various dimensions of the exposure
24. e PD model to be used for calculating the Conditional PD Credit Metrics Structural Model CMSM or Monte Carlo Expectation Maximization MCEM Methods are used LGD Models input the PD model to be used for calculating the Conditional LGD Linear Regression ML Linear Regression REML methods are used to predict LGD with help of forecasted values from simulation EAD Measure input the measure name as EAD is stored as a measure in OFSAAI PD Model Mitigants input the PD model used for calculating the Conditional PD for mitigants The models are the same as the Exposure PD models Future Value the measure storing the future value of the mitigant is selected The future value is calculated using Distribution Fitting VaR Percentile the confidence interval at which the Value at Risk is calculated Oracle Financial Software Services Confidential Restricted 28 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Teciraue Figure 22 Variable Selection for Outputs Output The output of the CDM model is VaR CVaR and Expected You need to select the measure corresponding to VaR CVaR EL and UL from the list in the variable name tab CDM loss model calculates the Loss LGD EAD with the probability given by PD value VaR Reader is used to calculate the VaR and CVaR with the given VaR percentile given as an input The following techn
25. e defining a Run In addition to the baseline runs simulation runs can be executed through the usage of the different Simulation Processes Such simulation runs are used to compare the resultant performance calculations with respect to the baseline runs This comparison will provide useful insights on the effect of anticipated changes to the business Run Definition A Run is a collection of processes that are required to be executed on the database The various components of a run definition are Process The user may select one or many End to End processes that need to be executed as part of the Run Run Condition When multiple processes are selected there is likelihood that the processes may contain rules T2Ts whose target entities are across multiple datasets When the selected processes contain Rules the target entities hierarchies which are common across the datasets are made available for defining Run Conditions When the selected processes contain T2Ts the hierarchies that are based on the underlying destination tables which are common across the datasets are made available for defining the Run Condition A Run Condition is defined as a filter on the available hierarchies Process Condition A further level of filter can be applied at the process level This is achieved through a mapping process Types of Runs Two types of runs can be defined namely Baseline Runs and Simulation Runs Baseline Runs are those bas
26. e is used for Mitigants Collateral or Nettable Liabilities It is used to predict the future value of the collaterals You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation Refer to the following screenshot The screen has two tabs Input and Output Variable Model Name Collateral FV Distribution Fitting Model Description Distribution fitting for future value collateral calculation Model Objective To calculate future value of collateral The modeling framework screens are as follows Oracle Financial Software Services Confidential Restricted 25 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Figure 20 Future Value Modeling Screen Input Inputs for future value distribution fitting are Simulation Model You need to select the simulated univariate model for predicting the collateral values Future Value Percentile the percentile as per which the future value multiple is computed Output Using the future value multiple the future collateral value is computed and stored as a measure which is used in the Conditional Default Model 3 2 8 Calculation of Conditional PD Conditional PD can be derived from any of the two approaches described as follows Credit Metrics Structured Model Monte Carlo Maximization Expect
27. ean Standard Deviation U Random Term that is uncorrelated with the variables Ov Standard Deviation of variables calculated from the variance covariance matrix Pvi v2 correlation between v1 and v2 calculated from variance covariance matrix For the above equations the coefficients Al A2 An An l are calculated A s are the coefficients of the general market or idiosyncratic variables Al A2 An An l and so on are substituted in the first equation to calculate Z Z and Unconditional PD is used to calculate Conditional PD For solving the above equations calculation of standard deviations correlation and variance covariance matrix of market variables is required Fe nore The intermediate statistical calculations like standard deviations variance covariance matrix and so on are calculated in the modeling framework It s automatically triggered when the credit metrics model is run You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation Refer the following screenshot You need to define the filters use the option of Time Referencing and Sampling for the variables as well Figure 23 Technique Selection The input and the output variables are defined by selecting the relevant variable in the modeling framework screen In this model general market variables affecting P
28. ed A Base Process Tree is a hierarchical collection of rules that are processed in the natural sequence of the tree The rules are sequenced in a manner required by the business condition The base process tree does not include sub processes that are created at run time during execution e A Simulation Process Tree as the name suggests is a tree constructed using a base process tree It is also a hierarchical collection of rules that are processed in the natural sequence of the tree It is however different from the base process tree in that it reflects a different business scenario The scenarios are built by either substituting an existing process with another or inserting a new process rules Introduction to Run In this chapter we will describe how the processes are combined together and defined as Run From a business perspective different Runs of the same set of processes may be required to satisfy different approaches to the underlying data Oracle Financial Software Services Confidential Restricted 42 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 The Run Framework enables the various Rules defined in the Rules Framework to be combined together as processes and executed as different Baseline Runs for different underlying approaches Different approaches are achieved through process definitions Further run level conditions or process level conditions can be specified whil
29. erarchical structure is adopted to facilitate the construction of a process tree A process tree can have many levels and one or many nodes within each level Sub processes are defined at level members and rules form the leaf members of the tree Through the definition of Process the user is permitted to logically group a collection of rules that pertain to a functional process Further the business may require simulating conditions under different business scenarios and evaluate the resultant calculations with respect to the baseline calculation Such simulations are done through the construction of Simulation Processes and Simulation Process trees Underlying metadata objects such as Rules T2T Definitions Non End to End Processes and Database Stored Procedures drive the Process functionality Types of Processes From a business perspective processes can be of 2 types End to End Process As the name suggests this process denotes functional completeness This process is ready for execution Non End to End Process This is a sub process that is a logical collection of rules It cannot be executed by itself It must be defined as a sub process in an end to end process to achieve a state ready for execution Process Definition A process is defined using existing rule metadata objects The various components of a process definition are Process Tree This is a hierarchical collection of rules that are processed in the natural
30. executed with a number of reports iterations up to 5000 for a set of 2000 records with a RAM of 16GB If you want to increase the number of records you should either decrease the number of reports or else increase the RAM size VaR Reader Normally a VaR model can be deployed in production infodom without even executing the same in the sandbox infodom But if transformations are used on the variables defined in the VaR model then you need to execute the same in the sandbox before he deploys that in the production infodom e Dummy Variables you need to classify the qualitative or categorical variables as dummy variables Transformations should not be applied on Dummy Variables while defining a model Oracle Financial Software Services Confidential Restricted 38 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Annexure A Terminologies and Analogies of the framework You must be familiar with the following terminologies that are constantly referred to in this manual Data Model A logical map that represents the inherent properties of the data independent of software hardware or machine performance considerations The data model consists of entities tables and attributes columns and shows data elements grouped into records as well as the association around those records Dataset It is the simplest of data warehouse schemas This schema resembles a star diagram While the center cont
31. flow model is only used for specialized lending This technique is used to calculate EAD for specialized lending exposures The process to define cash flow model is provided in the Technique 6 cash flow model to calculate UCPD under the heading Calculation of Unconditional PD EAD is the exposure amount if the net cash flow is less than the default threshold amount of that bucket These calculations are done in the Run Rule framework of the modeling framework GL Based and User Input for EAD Calculation If EAD is given as a user input or is GL based then the EAD amount is stored for its respective exposure Economic Capital is calculated for exposures sub exposures and mitigants The following section explains the Mitigants and Collateral Modeling 3 2 7 Mitigants or Collateral Modeling Mitigant Modeling is of two types in the Economic Capital EC product Provider Driven Mitigant Modeling Value Driven Mitigant Modeling or Collateral Modeling Provider Driven Mitigant Modeling is done by using Monte Carlo Expectation Maximization MCEM technique MCEM model is the same as PD Modeling for corporate sovereign bank and retail asset class The PD value is either calculated by MCEM technique or the PD value is given as download by the bank Value Driven Mitigant Modeling Collateral is done by Future Value Model Distribution Fitting e Technique 1 Future Value Collateral Distribution Fitting Objective This is a techniqu
32. he process parameter Remember the version of that model that you plan to select there At this stage select a relevant DT to populate the data outputs back to the fact Tables in the run immediately after each model The model and the accompanying DT together form a sub process Two DT s have been defined to populate data from REV_MODEL_OUTPUT_DETAILS to the FACT table FN_DT_UPD_FCT_PROD function is used to populate data for all the models to the fact table Update MCEM_PD function is used to populate PD values using the transformation e 1 e to the fact table from REV_MODEL_OUTPUT_DETAILS Missing Value Transformation should necessarily be applied on the general market variables while defining a Simulation model Missing Value should also necessarily be applied on the independent variables selected in Linear Regression or MCEM Models e Filters if you want to choose only a certain category of records in a model do this through Filters For example select asset class as Corporate to apply a CDM model only on Corporate In Future Value modeling one model is defined for each type of collateral For example For Gold one model is defined for Cash another model has to be defined Gold and Cash collateral types can be selected as filters in the FV models Oracle Financial Software Services Confidential Restricted 37 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 In Merto
33. inancial Services Economic Capital Advanced Release 1 1 2 0 0 where the product type is specialized lending securitization SFT OTC or where the customer type is Retail or the exposure amount is drawn ALL asset classes where the EAD GL Balance as downloaded exposure amount is drawn updated through Rules except where the product type is specialized lending securitization or where the customer type is Retail Retail asset class except EAD Pool Historical Average where the product type is SFT OTC or the exposure amount is drawn Retail asset class where the EAD GL Balance exposure amount is drawn ALL asset classes where the EAD VaR based product is SFT OTC Specialized Lending Corporate Sovereign Bank Conditional PD Credit Metrics Structural Model ne REP default rate Corporate Sovereign Bank Conditional LGD Linear Regression ML or REML Retail Conditional LGD Linear Regression ML or REML Historical Scenario based pool observed LGD Securitization Historical Loss Distribution Distribution Fitting Securitization SFA ULP Approach Distribution fitting Corporate Sovereign Bank Loss Calculation Conditional Default Model Average Distribution Fitting Banking Book Equity Loss Calculation VaR Reader ALL asset classes where the Loss Calculation Time To Default Model product type is OTC SFT Table 2 Calculation of Risk Factors 3 2 4 Calculation of Unconditional PD Unconditional P
34. inition Variable Selection Varabe O Figure 7 Technique Browser Screen You can define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation The input and the output variables are defined by selecting the relevant variable in the modeling framework screen Model Name Logistic Regression Model Description Regression for calculation of coefficients of General Market Variables Model Objective To calculate Unconditional PD You can define the filters hierarchies are used as filter use the option of Time Referencing and Sampling for the variables In logistic regression independent variables and dependant variables for the regression model to be run are provided as an input The measures defined for dependant independent variable is selected in the variable name tab drop down list Whether a particular variable is qualitative or quantitative should also be specified by you in the following screen Oracle Financial Software Services Confidential Restricted 13 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Modes Detass Mose namer Mosel Descroten Techeique t Pa tion Sec Exposutes kr Modeling 1 Cara tose font tening uosa Varaste TmeReterenceg fer Sampieg pus Process Outputs Osta Outputs Variable Selection f Venda
35. into the User Guide The topics in this section are organized as follows Scope of the Guide Audience Document Conventions Scope of the Guide This manual aims at assisting the user use the Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Product The manual elaborates the function process flows and the techniques used in the computation of EC Audience This manual is designed for Application System Administrators Business Analysts Strategists and Bankers for the purpose of calculating and analyzing the historical data and predicting the defaulters Document Conventions Certain practices have been incorporated into this document to help you easily navigate through the document The table given below lists some of the document conventions incorporated into this User Guide Conventions Description Bold User Interface Terms Italics Keywords Emphasis e Cross References Table 1 Document Conventions The other document conventions incorporated into this User Guide are as follows In this document a Note is represented as shown below Important or useful information has been represented as a Note Oracle Financial Software Services Confidential Restricted iii User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 1 Introduction Oracle Financial Services Analytical Application Infrastructure OFSAAJ is an analytical application platf
36. ique and a relevant data set to be used in the calculation Model Name Linear Regression Model Description Linear Regression for EAD calculation Model Objective EAD You can define the filters hierarchies are used as filter using the option of Time Referencing and Sampling for the variables The dependent variable in this technique is the drawdown percent data output is CCF percent and the independent variables are the general market and idiosyncratic variables The input or the output variables are defined by selecting the relevant variable in the modeling framework screen The measures defined for dependant independent variable is selected in the variable name tab drop down list Whether a particular variable is qualitative or quantitative should also be specified by you The following screen show how the models are defined in modeling framework Oracle Financial Software Services Confidential Restricted 23 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Model Detass mosa namer LAD Las Mosel Descreton Techesue Loex Reyers Ey Caltrata Wesel Variable TmeReterencng fter Sampleg puts Process f Variable Selechon Figure 19 Variable Selection for EAD calculation Variables affecting all exposure sets are provided as an input Sets are formed such that for all exposures its drawdown percent is affected by the same set of variables In addition t
37. ique comes into picture when the unconditional PD is calculated by Merton Model or Credit Metrics Transition Matrix Model You need to specify the model ID of the credit matrix structural model as an input in the conditional default model If unconditional PD is calculated by MCEM then the following technique is not used in the conditional default model to calculate conditional PD Technique 1 Credit Metrics Structural Model This technique is used in the calculation of conditional PD This technique is not run independently but is run in the Conditional Default Model Credit Metrics is a framework used for quantifying credit risk portfolios credit products and market driven instruments In EC product Credit Metrics Structural Model is used to calculate the Conditional PD for an exposure The banks provide the percentage effect of idiosyncratic variables on the counterparty The bank also provides by what percentage the counterparty is affected by a particular general market variables Objective The objective of the model is to compute the conditional probability of default by solving the following equations Z A1VIU A2V2 AnVn A nU Ai 1 Y ZA Ovi Oe Zi Vi An 1 1 1 Oracle Financial Software Services Confidential Restricted 29 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Where n number of factors for the specified counterparty V Standard Value which is equal to V M
38. like its customer type product type and so on Risk weight is an example of a business processor Oracle Financial Software Services Confidential Restricted 43 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Why Define a Business Processor Measurements that require complex transformations that entail transforming data based on a function of available base measures require business processors A supervisory requirement necessitates the definition of such complex transformations with available metadata constructs Business Processors are metadata constructs that are used in the definition of such complex rules Refer to the section Accessing Rule in the Rules Process and Run Framework Manual for more details on the use of business processors Business Processors are designed to update a measure with another computed value When a rule that is defined with a business processor is processed the newly computed value is updated on the defined target Let us take the example cited in the above section where risk weight is the business processor A business processor is used in a rule definition Refer to the section defining a Rule in the Rules Process and Run Framework Manual for more details In this example a rule is used to assign a risk weight to an exposure with a certain combination of dimensions Oracle Financial Software Services Confidential Restricted 44 User Guide Oracle Financial Services Econ
39. n Model a negative value for Distance to Default is possible Two rules are defined to populate the PD value from the stage table to the fact table Non Sec Mitigant Unconditional PD Merton is used to populate normal values PD to the fact table where as Non Sec Mitigant Unconditional PD Merton Negative DD rule is used to update the unconditional PD column to 1 in the fact table If you opt to use Regression method to calculate PD for Retail Asset Class then you shall use Linear Regression ML or Linear Regression REML to calculate PD unlike logistic regression which is used to calculate PD for other asset class like corporate sovereign and bank Variable Transformations should be appropriately handled in a model For example in Linear Regression Model the pre model and post model transformations should be defined on the dependant variable and data output respectively If pre model Transformation is InvNorms Beta Y then the post model transformation should be BetaInv Norms Y e LGD Model if you want to define an LGD model with Linear Regression do the InvNorms Beta Y transformation on the dependant variable While doing so you should remember to apply Outlier Detection transformation before the InvNorms Beta Y transformation on the dependant variable with Upper Bound Value 0 9999 and Lower Bound Value 0 0001 e Monte Carlo Expectation Maximization During model fitting inside the sandbox MCEM can be
40. nal LGD cccccccsssscccesssecesssceceessesecsessecsesssecsessessssuseecsessssecessessesuseecsesssseceaaes 20 3 2 6 Calculation Of EAD iS iia 22 3 2 7 Mitigants or Collateral MOGeHIG a aa ii init aia AEEA A Aa As Saa EREA as Sants 25 3 2 8 Calculation of Conditional PD u ccsccccccssccesssssccessscesssececeessssecessesessssecesssessssussecsessseesssesscassecsesssseceaaes 26 3 2 9 Calculation of Conditional LGD ccccccccesssssecessssesssscecsessssecessecessssecsessessesseecsessscesssesscusseceesssseensaes 26 BZ TO LOSS MOGELCOICUIGTION A A A A E a AA AA its 27 32 I MODERN EXECUTION a A a a atado 34 3 22 12 MOOD Me E aa a dances cohesdandanedae sondeads nines nde onddaslechesbendunsliaeens 35 32 133 RUN DEFINAN da 36 32 145 RUN EXCCUCIO asii it id A a iaa add dd tica vids iaa dotes id 36 32 15 Important Notes aceia at aetate aata a eaa bee aaa aiaa iaa eiat AAAA iiia iaaa 36 3 2 16 Things t REMeMDer iii aai sexes a a E a E E TEn 37 ANNEXURE A TERMINOLOGIES AND ANALOGIES OF THE FRAMEWORK 4 39 ANNEXURE B GENERATING DOWNLOAD SPECIFICATIONS sseesseesssosesoseeeseesssoseeeseeesee 45 GLOSSARY 5555555 SSR RNa saa ee e e e aoaeeoo 46 Oracle Financial Software Services Confidential Restricted ii User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 About the Guide This section provides a brief description of the scope the audience and document conventions incorporated
41. nconditional LGD Model Objective Linear Regression To calculate Unconditional LGD Oracle Financial Software Services Confidential Restricted 20 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 The following screens show how the models are defined in modeling framework Model Definition Model Delete Mosel Name Mosel Desereeen Technique Callate Mosel Techniave Variable Teme Meterencieg Fiet o Hechos Variable Selection AT vroen Figure 17 Technique Selection for LGD You can define the filters hierarchies are used as filter using the option of Time Referencing and Sampling for the variables In linear regression model definition independent variables and dependant variables for the regression model to be run are given as input The measures defined for dependant independent variable is selected in the variable name tab drop down list Process Output Modeling framework supports the following process outputs for logistic regression 2 log likelihood Beta Coefficients Gamma Values Number of fixed effects estimated Number of random effects estimated Standard Error t value The process outputs for both REML and ML linear regression is the same Whether a particular variable is qualitative or quantitative is also specified by you in the following screen Oracle Financial Software Services Confidential Restricted 21 User Guide
42. o all the variables affecting PD LGD original maturity bucket and residual maturity bucket affects the EAD calculation Process Outputs Process outputs for the model are shown below The technique gives standard linear regression outputs Beta Coefficients Covariance Error degrees of freedom F Statistics R Square Statistics Residual Mean Sum of Squares Residuals Standard Error p value You can choose any or all of the above parameters as process outputs The process outputs calculated are the coefficients of each exposure set for that particular date Goodness of Fit parameters like residuals residual mean sum of squares R square statistics and F statistics is calculated Parameter statistics like standard error and p value are also calculated Model Outputs All the parameters are stored in model output screen You have to note down either model Oracle Financial Software Services Confidential Restricted 24 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 objective or model code to view the model output Technique 2 Pool Historical Average This technique is used for retail asset class only Historical data like drawdown percent and undrawn exposure amount is provided as download by the bank to calculate EAD Pool Historical Average EAD Average of historical drawdown undrawn amount e Technique 3 Cash Flow Model for EAD Calculation Cash
43. of each exposure set for a particular date Goodness of Fit parameters like residuals is calculated Model implementation statistics odds ratio and parameter statistics like Wald statistics t statistics standard error and p value are also calculated Classification parameters like classification table model reports and so on are given as output Model Outputs All the parameters are stored in the model output screen You have to note down either the model objective or model code to view the model output Corporate Sovereign Bank asset class uses the above mentioned techniques to calculate Unconditional PD Retail also uses Linear Regression but does not use Credit Metrics Transition Matrix and Merton Model In addition to linear regression retail asset class use Historical Average Pool Default Rate for calculation of unconditional PD e Technique 5 Historical Average Pool Default Rate This method is only used for Retail Asset Class The probability of default PD of the retail pool is calculated as the mean of the default rates of the pool for all the historical dates which is given as a download by the bank The calculation of average default rate is done with the help of a Type III rule Retail also uses the Linear Regression technique to calculate the PD as an alternative to this method e Technique 6 Cash Flow Model Objective Cash Flow model is used to calculate PD LGD and EAD for specialized lending This model is u
44. omic Capital Advanced Release 1 1 2 0 0 ANNEXURE B Generating Download Specifications Data Model for OFS Economic Capital Advance Release is available on customer request as an ERwin file Download Specifications can be extracted from this model Refer the whitepaper present in OTN for more details Oracle Financial Software Services Confidential Restricted 45 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Glossary Unconditional Probability of Default Probability of a customer defaulting one holding period later given the information available today The holding period is generally assumed to be one year Conditional Probability of Default Probability of a customer defaulting one holding period later given a scenario for information that will become available in future The holding period is generally assumed to be one year Unconditional Loss Given Default The loss anticipated if the customer happens to default one holding period later given the information available today The holding period is generally assumed to be one year Conditional Loss Given Default Loss anticipated if the customer happens to default one holding period later given a scenario for information that will become available in future The holding period is generally assumed to be one year VaR based Economic Capital Credit Risk A user chosen quantile of the loss distribution of the customer one holding period later The
45. or Business Processors A measure refers to the underlying column value in the database and you can consider this as the direct value available for modeling You can select hierarchy for modeling purposes For modeling purposes qualitative variables need to be converted to dummy variables and such dummy variable needs to be used in Model definition Dummy variables can be created on a hierarchy Business Processors are used to derive any variable value You can include such derived variables in model creation 3 2 3 Model Definition This section deals with models to be defined in the modeler part of the product Models are defined for calculation of risk measures PD LGD and EAD according to each Asset Class Economic Capital Product calculates PD LGD and EAD for the Basel II prescribed Asset Class All model definitions are done on a set of exposures affected by the same set of variables Dataset should be selected specific to the model defined Data set filter is used to group the exposures into one set Models are defined for each technique to calculate risk factors Dataset relevant to the particular technique needs to be selected You can define the input and the output variables for the technique the variables defined in the sandbox are shown in the dropdown list can add filters and define data and process outputs by selecting the respective tabs as shown in the following screen Oracle Financial Software Services Confidential Rest
46. or dependant independent variable is selected in the variable name tab drop down list Whether a particular variable is random or fixed is also specified by you in the following screen Oracle Financial Software Services Confidential Restricted 10 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Mode Definition _ Model Detada Uode Name Medal Desergnen Technique ll Coenie Mote Varatia Tee Reterencng _ Variable Selection AT vare o SANATSAN Figure 5 Adding Variable To run MCEM you will have to specify two parameters Parameters e Number of Reports The number of times each record is blown up to calculate the coefficients For example If there are 50 records and you provide 7 as the number of reports then one record is blown up to 7 simulated values Hence 50 records are blown up to 50 7 350 records Omissions the number of iterations of the random error to be omitted before the calculation of coefficients Process Output Beta Coefficients Beta Values for unsuccessful iterations Residuals Standard Error Wald Statistics Odds Ratio p value Oracle Financial Software Services Confidential Restricted 11 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Mode Definition Variable Tere Referencing Fae Sameing byuis Process Outruts Osis Ostpsts Di Figure 6 Process outputs Screen Yo
47. orm which has been architected to be multi tiered and open systems compliant OFSAATI is fully web enabled It s a 100 thin client browser based interface with zero foot print which dramatically reduces the cost of application deployment All OFSAAI processes including those related to business are metadata driven thereby providing a high degree of operational and usage flexibility and a single consistent view of information to all users OFSAAI product suite includes a rules framework designer engine Unified Metadata Manager which has a semantic layer of metadata abstraction that is common over both relational and OLAP repositories 1 1 Analytical Applications Overview Analytical Applications for instance Basel II Economic Capital and so on are pre packaged on OFSAAI and are ready to install Regulatory capital planning has been the most important aspect for a bank s risk compliance activity Basel II Framework promotes adoption of stronger risk management practices by banks which will address all major risks comprehensively In addition to regulatory capital economic capital has become a frequent concept used in the analysis of the new framework on bank capital regulations prescribed by the Basel II committee Basel II introduces economic capital into the regulatory capital consideration by requiring banks to determine capital adequacy based on the risk of specific businesses Risk is the measurement of unexpected loss which ari
48. ows for the following approaches Linear Regression based estimates of CCFs Pool Historical Average Only Retail Cash flow based for Specialized Lending e GL Based Only Drawn Oracle Financial Software Services Confidential Restricted 22 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 EAD as User Input Drawn product type uses GL based methods for calculation of EAD SFT and OTC use time to default model Among asset class Specialized Lending uses Cash Flow Model and Retail uses Pool Historical Average Method to calculate EAD Technique 1 Linear Regression for EAD Calculation This technique is used for all asset classes except specialized lending retail and securitization All other product types in corporate sovereign bank uses regression based estimate of CCF to calculate EAD except for the product types drawn SFT s and OTC s Objective The objective of the model is to calculate the EAD by Simple Linear Regression without any fixed and random effects Historical Values of all the factors or variables at time t 1 and historical values of credit conversion factor CCF at time t are provided as a download User inputs are the General Market Variable s and Idiosyncratic Variable s affecting the exposures and the corresponding random error terms You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the techn
49. ows the model definition screen for the cash flow model Model Definition Figure 16 Cash Flow Model Definition Model Name Cash Flow Model Oracle Financial Software Services Confidential Restricted 19 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Description Cash Flow Model for specialized Lending Model Objective To calculate Unconditional PD for specialized lending Select the appropriate dataset from the drop down list On selecting the data set the cash flow entries updated in the data entry screen relevant to the dataset appears in the cash flow entries textbox Select the Simulation Model ID from the drop down list The simulation models are defined before the model definition or data entry of cash flow model Process or Data Outputs The Net Cash flow Inflow Outflows is calculated in the modeling framework Default Definitions and Loss is calculated with the help of the Run Rule framework Depending on the default status Unconditional PD is calculated Model Outputs All the parameters are stored in the model output screen You have to note down either the model objective or model code to view the model output 3 2 5 Calculation of Unconditional LGD Unconditional Probability of Default can be derived based on any of the following 3 approaches Linear Regression Historical Default Weighted average of pool observed LGD only Retail
50. ricted 4 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition Figure 1 Model Definition The computation of unconditional PD conditional PD unconditional LGD conditional LGD and EAD are done at a set level The inputs and outputs of the techniques and defining model for each technique are explained as follows Before model definition and variable preparation the Asset Class Reclassification needs to be done The bank gives exposure level information as a download The exposure level information is either reclassified according to the Basel rules or according to the Banks own reclassification rule The following table displays the techniques used in the calculation of risk factors for each asset class Corporate Sovereign Bank Unconditional PD Credit Metrics Transition Matrix Merton Model Logistic Regression MCEM Retail Unconditional PD Logistic Regression MCEM Historical Average Pool Default Rate Specialized Lending Unconditional PD Cash flow based Corporate Sovereign Bank Unconditional LGD Linear Regression ML or REML Retail Retail Unconditional LGD Linear Regression ML or REML Historical Default weighted average of pool observed LGD Specialized Lending Unconditional LGD Cash Flow based ALL asset classes except Simple Linear Regression Oracle Financial Software Services Confidential Restricted 5 User Guide Oracle F
51. ritization Banking Book Other Assets Credit Risk modeling revolves around three primary entities Exposure which is an account or a pool forms the basic entity at which credit risk modeling would be done had there been no mitigants Mitigants which are mapped to exposures in an m n fashion and can be used to reduce the risk due to an exposure Sub exposure which is defined as a unique exposure mitigant combination which is either fully covered by a mitigant or uncovered Sub exposures are created in a manner which minimizes regulatory capital Mitigants are divided into two categories Provider default driven that is guarantees and credit derivatives Value driven like collateral nettable deposits and so on Economic Capital needs to be computed for exposures sub exposures and in some situations for provider driven mitigants While calculating Economic Capital EC for exposures mitigants are assumed to be redundant 3 1 Scope of Oracle Financial Services Economic Capital Advanced Oracle Financial Services Economic Capital Solution Release 1 1 2 0 0 computes undiversified Economic Capital EC for Credit Risk individually for each exposure in the portfolio and the portfolio EC will be the sum of individual components This methodology ignores diversification but at the same time allows the bank to attribute risk to individual lines of business without having to go through an allocation process RAROC and Sharehol
52. robability of Default PD can be derived based on any of the following approaches e Technique 1 Credit Metrics Transition Matrix A rating based Transition Matrix for the counterparty and its present rating is provided as a download The unconditional default probability is directly read from the relevant cell of the Oracle Financial Software Services Confidential Restricted 6 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 transition matrix In this method provide input parameters such as what asset class the counterparty belongs to source of rating current rating of the counterparty The rating given should be from one source at a specified time Transition Matrix and Rules Framework is used for the computation of PD A Type IIT rule updates the probability of default PD column in Fct_Non_Sec_Exposures Technique 2 Merton Model This method mainly deals with iteratively solving four simultaneous equations called Black Scholes equations You need to provide with the daily prices of the equity of the counterparty from which to calculate the market price of the bank s assets and Distance to Default by solving the Black Scholes equations Objective The objective of the model is to compute the probability of default by solving the Black Scholes model Variables E F and r are given as a download and oE is calculated in the modeling framework V and oV are populated into the customer table using T
53. rom a computation performed on a simple measure The computation that is performed on the measure often involves the use of statistical mathematical or database functions Oracle Financial Software Services Confidential Restricted 39 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Modeling Framework The OFSS Modeling Environment performs estimations for a given input variable using historical data It relies on pre built statistical applications to build models The framework stores these applications so that models can be built easily by business users The metadata abstraction layer is actively used in the definition of models Underlying metadata objects such as Measures Hierarchies and Datasets are used along with statistical techniques in the definition of models Introduction to Rules Institutions in the financial sector may require constant monitoring and measurement of risk in order to conform to prevalent regulatory amp supervisory standards Such measurement often entails significant computations and validations with historical data Data must be transformed to support such measurements and calculations The data transformation is achieved through a set of defined tules The Rules option in the Rules Framework Designer provides a framework that facilitates the definition and maintenance of a transformation The metadata abstraction layer is actively used in the definition of rules where the u
54. s values of DD and UC PD applicable for all counterparties is given as a download Model Outputs Distance to default is stored in model output screen You have to note down either the model objective or model code to view the model output e Technique 3 Monte Carlo Expectation Maximization MCEM Economic Capital Product uses Monte Carlo Expectation Maximization MCEM to calculate unconditional probability of default MCEM method uses logistic regression to predict the values of the coefficients MCEM calculates the coefficients of the variables iteratively Objective The objective of the model is to calculate unconditional PD with statistical techniques like Regression and Monte Carlo Expectation Maximization MCEM MCEM calculates the coefficients of the general market variables by iteration used in the logistic regression equation It takes into account the random effect of the variables MCEM is a work about for logistic regression with random effects You need to provide the following inputs X Historical values of all the factors at time t Y Exposure Defaulted Position that is Exposure has Defaulted or not Yes No at time t 1 All exposure and model mapping for PD All guarantee and mitigants mapping for PD The regression equation for used in MCEM is Y aygrta X 99 X gt apX tE Eot En Where Y 1n PD 1 PDb ao intercept a a2 an coefficients of the variables X s X 1 X g
55. sed in situations where default rather than being based on counterparty solvency is driven by a set of cash flows You can enter the data in the data entry screen before the cash flow model is defined The data entry screen is located under the Application tab in the left hand side LHS of the modeling framework screen The highlighted tab to the left shows the flow to reach the Data Entry Screen Refer to the following screenshot for more information on data entry Oracle Financial Software Services Confidential Restricted 15 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Definition 2 Modei Cetass Loss Fatmatos tor Specaived Lenang Loss Estimates sar Specaizes tenang Mose Name Econcme Cantal Cash Flow Model Speciatzed Lending Dataset fee rous A Parameters pr Confidence Level CasaFiow Equations Senutaton Mode Figure 10 Model Definition Model Definition Medes Wansremert gt Model Deaden Lot Mode 2 Model Detats uosa namer Loss Lstmatos tor Specaized Lenses N Losa Ealerettes te Egecmices Mosel Deserenon Leading Econcmic Capta Vecheique Scecatzed Lencng Dataset Fite bps Variable Data Outpsta _ 2 Variable Selection j oth de El Vante a AIMA Lode GOP Growin Fate Forecasting 0 user into Created By ISFEB ION 125259 PU Figure 11 Model Definition Oracle Financial Software Services Confidential Restricted 16 User Guide Ora
56. sequence of the tree The process tree can have levels and members Each level constitutes a sub process Each member can either be a Type 2 rule or Type 3 rule an existing non end to end process a Type 1 rule T2T or an existing transformation that is defined through Data Integrator If no predecessor is defined the process tree is executed in its natural hierarchical sequence as explained in the stated example Oracle Financial Software Services Confidential Restricted 41 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Example In the above figure first the sub process SP1 will be executed The sub process SP1 will be executed in following manner Rule 1 gt SPla gt Rule 2 gt SP1 The execution sequence will be start with Rule 1 followed by sub process SPla followed by Rule 2 and will end with sub process SP1 The Sub Process SP2 will be executed after execution of SP1 SP2 will be executed in following manner Rule 3 gt SP2 The execution sequence will start with Rule 3 followed by sub process SP2 After execution of sub process SP2 Rule 4 will be executed and then finally the Rule 5 will be executed The Process tree can be built by adding one or more members called Process Nodes If there are Predecessor Tasks associated with any member the tasks defined as predecessors will precede the execution of that member Type of Process Trees Two types of process trees can be defin
57. ser is permitted to re classify the attributes in the data warehouse model thus transforming the data Underlying metadata objects such as Hierarchies that are non large or non list Datasets and Business Processors drive the Rule functionality Types of Rules From a business perspective Rules can be of 3 types Type 1 This type of Rule involves the creation of a subset of records from a given set of records in the data model based on certain filters This process may or may not involve transformations or aggregation or both Such type 1 rule definitions are achieved through Table to Table T2T Extract Refer to the section Defining Extracts in the Data Integrator User Manual for more details on T2T Extraction Type 2 This type of Rule involves re classification of records in a table in the data model based on criteria that include complex Group By clauses amp Sub Queries within the tables Type 3 This type of Rule involves computation of a new value metric based on a simple measure and updating an identified set of records within the data model with the computed value Rule Definition A rule is defined using existing metadata objects The various components of a rule definition are Dataset This is a set of tables that are joined together by keys A dataset must have at least one FACT table Type 3 rule definitions may be based on datasets that contain more than 1 FACT tables Type 2 rule definitions must be base
58. ses because of more than anticipated liability or less than expected returns For example unexpected losses due to unexpected fluctuations in the market prices Unexpected loss can arise due to any risk type Basel encourages measurement of economic capital so that an aggregated capital is calculated by aggregating all the risks together with one single risk metrics With the proposal of new capital accord understanding and measuring economic capital has become a compliance obligation Economic Capital is the capital level required by the bank to cover the losses with a given probability Economic capital is attributed mainly by three risks Credit Risk Operational Risk and Market Risk Economic Capital methodologies can be applied across products lines of business and other segments as required Basel Committee emphasizes banks to have their own methods processes to calculate adequate capital for the risk they assumed The focus on Economic Capital calculation is industry wide to measure risk to optimize performance by reallocating capital to strategically important businesses as well as businesses with high returns The Scope of Economic Capital Advanced Analytics is limited to Economic Capital Credit Risk Oracle Financial Software Services Confidential Restricted 1 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 2 Installing the Oracle Financial Services Economic Capital Advanced Once the OFSAAI Infra
59. structure has been loaded the Oracle Financial Services Economic Capital Advanced Application has to be installed To install Oracle Financial Services Economic Capital Advanced Application refer to the Installation Manual 2 1 Model Upload To carry out the Model Upload click Unified Metadata Manager on the left pane of the OFSAAI Infrastructure Under that click Import Model to open the Business Model Upload screen Choose the type of Upload as New Upload Enter the Erwin XML File Path and click Upload and the model will be uploaded Oracle Financial Software Services Confidential Restricted 2 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 3 Oracle Financial Services Economic Capital Advanced Economic Capital EC is the capital that banks and other financial institutions hold to insure debt holders in general and depositors in particular against potential loss scenarios over a given time period typically a year An Economic Capital framework relates the risk of investment to the amount of capital required thus putting business performance or investment return into a proper perspective relative to risk On identifying the risk type EC is computed according to the asset class of the counterparty Different approaches are followed for counterparties falling under different asset class The asset type prescribed according to Basel is Corporate Sovereign Retail Specialized Lending Secu
60. t X General Market Idiosyncratic Variables at time t 1 Random Error Terms related to X s X s can be General Market Variables or Idiosyncratic Variables or both Based on the variables affecting each exposure different subsets are formed Each subset should have the same set of affecting variables Regression is run on each subset and a regression equation is fetched for each subset Oracle Financial Software Services Confidential Restricted 9 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 The following figure shows the definition of Monte Carlo Expectation Maximization in the modeling framework a Model Definition Figure 4 Technique Selection in Model You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation The input and the output variables are defined by selecting the relevant variable in the modeling framework screen Model Name Monte Carlo Expectation Maximization Model Description MCEM for calculation of coefficients Model Objective To calculate Unconditional PD You can define the filters hierarchies are used as filter use the option of Time Referencing and Sampling for the variables In MCEM independent variables and dependant variables are given as an input The measures defined f
61. te predicted loss percentages using the simulation model Objective The objective of CDM is to calculate the risk factors like Conditional PD Conditional LGD EAD VaR EL and UL You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation Refer to the following screenshot Technique selected is Conditional_Default_Model which is one of the business models in the product Dataset selected is Sec_Exposure_Modelling The Securitization and Non Securitization modeling is done in different sandbox The following screen shows the filters to be used input and output variables Model Definition Figure 25 Distribution Fitting Model for Securitization Input Inputs of the CDM Distribution Fitting model are Simulation Model select the simulation model that generates loss percentages according to the distribution fitted The simulation model is a univariate simulation model EAD Measure The measure which stores the exposure at default is selected from the list of measures EAD is given as a download Oracle Financial Software Services Confidential Restricted 32 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 VaR Percentile VaR percentile is provided as an input by you Enter the percentile at which VaR should be calculated Output The output
62. u can choose any or all of the above parameters as process outputs The process outputs calculated are the coefficients of each exposure set for a particular date Goodness of Fit parameters like standard error and residuals are calculated Model implementation statistics odds ratio and parameter statistics like Wald statistics and p value are also calculated Model Outputs All the parameters are stored in the model output screen You have to note down either the model objective or model code to view the model output e Technique 4 Simple Logistic Regression Generalized Linear Mixed Models Generalized Linear Mixed Model GLMM is the 3 method you can choose from to compute PD GLMM consists of two types of regression Linear Regression and Logistic Regression The second step to calculate unconditional PD is defining model for logistic regression You can choose Simple Logistic Regression for the calculation of unconditional PD but MCEM is preferable as it considers the randomness of variables The equation for regression is the same as mentioned in MCEM Objective The objective of the model is to calculate the unconditional PD by Simple Logistic Regression without any fixed and random effect The following screens show how the models are defined in modeling framework Oracle Financial Software Services Confidential Restricted 12 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Model Def
63. ulated by Monte Carlo Expectation Maximization Model You need to specify the model ID of the MCEM as an input in the conditional default model The coefficients of the variables affecting the PD are fetched and the prediction is run to calculate the conditional PD The simulated values of the variables are substituted in the regression equation and the dependant variable logit function Y is calculated Y is transformed to give the conditional PD As there are 10 000 scenarios for each exposure 10k values of conditional PD is Oracle Financial Software Services Confidential Restricted 31 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 calculated For Conditional LGD calculation the Beta coefficients are substituted in the regression equation with the simulated values of the variables We get 10 000 values of LGD these values are got after post transformations in the model e Loss Model 2 Conditional Default Model Distribution Fitting Conditional Default Model Distribution Fitting is used only for Securitization Before this model is defined in the sandbox the pre requisite models should have been defined deployed and executed The pre requisite models are Distribution Fitting Model Simulation Model The Distribution Fitting model is run on the historical loss percentages The model gives two parameters of the distribution mean and standard deviation Mean and Standard Deviation is used to simula
64. ype III rule Other variables are stored in the parameters table E VO d1 esTFD d d1 In V F 0 50 T 6 WT d2 d GyWT p V ED dp Ov Where og Weekly Counterparty Equity returns standard deviation E Current market price of Equity of the above counterparty F Debt amount of the Counterparty and d1 d2 are the intermediate variables V market price of the bank s assets Oy Standard deviation of bank s assets standard normal CDF function You need to define the Model Name Model Description and Model Objective in the modeling framework The screen asks for the technique and a relevant data set to be used in the calculation Refer to the following screenshot The input or the output variables are defined by selecting the relevant variable in the modeling framework screen You can define the filters using Time Referencing and Sampling for the variables as well Oracle Financial Software Services Confidential Restricted 7 User Guide Oracle Financial Services Economic Capital Advanced Release 1 1 2 0 0 Mode Definition _ Modei Detada ose Nare Caderpaty PO Eatrratce Corporate We Tha mode a used t estrate Matai Deecreten Proabaty of Detaut for Corporate Economie Copan Techeigue Merton Vode Non Sec Exposutes for Modeling Varete Time Reference file Sempleg puts Osta Outputs Parameters Risk Pree Rate Tre Figure 2 Model Definition Screen In this

Download Pdf Manuals

Related Search

Related Contents

EXSYS EX-1183HMVS-W  MTX Audio MTX Thunder 4405 User's Manual  "取扱説明書"  Roadstar CDR-4200CD/BK CD radio  取扱説明書 活用ガイド  

Copyright © All rights reserved. Failed to retrieve file