1| Page Basel II and Basel III Credit, Market, Operational, and Liquidity Risks with Asset Liability Management It is often said that the Basel Committee Standards, formally called Capital Accords, constitute the bible for banking regulators (Central Banks) everywhere. In addition to the Accords, the Basel Committee has also framed 29 principles for effective banking supervision known as the Core Principles for Effective Banking Supervision. These standards encompassed by the Capital Accord and the Core Principles have become the source of banking regulation in every country in the world. As is widely known, these standards have evolved from Basel I to Basel II and III, reflecting the evolution of the financial industry (from Basel I to II) and the lessons from the financial crisis of 2008 (from Basel II to III). The most noticeable financial regulation paradigm changes captured and fostered by the Basel standards’ evolution are risk management and capital allocation. These most important changes in the international standards, and, therefore, in virtually every country´s financial regulatory framework, relate to the manner in which risks are managed and capital is calculated. By the general definition, as stated in Core Principle 15, Risk Management is the process to be used to “identify, measure, evaluate, monitor, report and control or mitigate all material risks on a timely basis and to assess the adequacy of their capital and liquidity in relation to their risk profile.” This process has been presented as the IMMM process: Identify, Measure, Monitor, and Mitigate each risk. In practice, the way to manage risks, and, hence, comply with the new Basel regulations, is to introduce or enhance the IMMM process for each material risk the financial institution faces. Along with the aforementioned international standards, there are tools that facilitate the implementation or enhancement of the IMMM processes. Briefly, these are (i) Formal Policies; (ii) Key Risk Indicators; (iii) Capital Models; and (iv) MIS/Reports. This case study looks at the practical tools—quantitative models, Monte Carlo risk simulations, credit models, and business statistics—utilized to model and quantify regulatory and economic capital, measure and monitor key risk indicators, and report all the obtained data in a clear and intuitive manner. It relates to the modeling and analysis of asset liability management, credit risk, market risk, operational risk, and liquidity risk for banks or financial institutions, allowing these firms to properly identify, assess, quantify, value, diversify, hedge, and generate periodic regulatory reports for supervisory authorities and Central Banks on their credit, market, and operational risk areas, as well as for internal risk audits, risk controls, and risk management purposes. In banking finance and financial services firms, economic capital is defined as the amount of risk capital, assessed on a realistic basis based on actual historical data, the bank or firm requires to cover the risks as a going concern, such as market risk, credit risk, liquidity risk, and operational risk. It is the amount of money that is needed to ensure survival in a worst‐case scenario. Financial services regulators such as Central Banks, Bank of International Settlements, and other regulatory commissions should then require banks to hold an amount of risk capital equal at least to its economic capital times some holding multiple. Typically, economic capital is calculated by determining the amount of capital that the firm needs to ensure that its realistic balance sheet stays solvent over a certain time period with a prespecified probability (e.g., usually defined as 99.00%). Therefore, economic capital is often calculated with Value at Risk (VaR) type models.
ItisoftensaidthattheBaselCommitteeStandards,formallycalledCapitalAccords,constitutethebible for banking regulators (Central Banks) everywhere. In addition to the Accords, the BaselCommittee has also framed 29 principles for effective banking supervision known as the CorePrinciples forEffectiveBankingSupervision.ThesestandardsencompassedbytheCapitalAccordand the Core Principles have become the source of banking regulation in every country in theworld.Asiswidelyknown,thesestandardshaveevolvedfromBaselItoBaselIIandIII,reflectingtheevolutionofthefinancialindustry(fromBaselItoII)andthelessonsfromthefinancialcrisisof2008(fromBaselIItoIII).ThemostnoticeablefinancialregulationparadigmchangescapturedandfosteredbytheBaselstandards’evolutionareriskmanagementandcapitalallocation.Thesemostimportant changes in the international standards, and, therefore, in virtually every country´sfinancial regulatory framework, relate to themanner inwhich risks aremanaged and capital iscalculated.Bythegeneraldefinition,asstatedinCorePrinciple15,RiskManagementistheprocesstobeusedto“identify,measure,evaluate,monitor,reportandcontrolormitigateallmaterialrisksonatimelybasisandtoassesstheadequacyof theircapitaland liquidity inrelationtotheirriskprofile.” This process has been presented as the IMMMprocess: Identify,Measure,Monitor, andMitigate each risk. In practice, theway tomanage risks, and, hence, complywith the newBaselregulations, is to introduce or enhance the IMMM process for each material risk the financialinstitutionfaces.
Along with the aforementioned international standards, there are tools that facilitate theimplementationorenhancementoftheIMMMprocesses.Briefly,theseare(i)FormalPolicies;(ii)KeyRiskIndicators;(iii)CapitalModels;and(iv)MIS/Reports.
Thiscasestudylooksatthepracticaltools—quantitativemodels,MonteCarlorisksimulations,credit models, and business statistics—utilized to model and quantify regulatory and economiccapital,measure andmonitor key risk indicators, and report all theobtaineddata in a clear andintuitivemanner.Itrelatestothemodelingandanalysisofassetliabilitymanagement,creditrisk,market risk, operational risk, and liquidity risk forbanksor financial institutions, allowing thesefirmstoproperlyidentify,assess,quantify,value,diversify,hedge,andgenerateperiodicregulatoryreportsforsupervisoryauthoritiesandCentralBanksontheircredit,market,andoperationalriskareas,aswellasforinternalriskaudits,riskcontrols,andriskmanagementpurposes.
Inbankingfinanceandfinancialservicesfirms,economiccapitalisdefinedastheamountofriskcapital, assessedon a realistic basisbasedon actual historical data, thebankor firm requires tocover therisksasagoingconcern,suchasmarketrisk,creditrisk, liquidityrisk,andoperationalrisk.Itistheamountofmoneythatisneededtoensuresurvivalinaworst‐casescenario.FinancialservicesregulatorssuchasCentralBanks,BankofInternationalSettlements,andotherregulatorycommissions should then require banks to hold an amount of risk capital equal at least to itseconomic capital times some holding multiple. Typically, economic capital is calculated bydeterminingtheamountofcapitalthatthefirmneedstoensurethatitsrealisticbalancesheetstayssolventoveracertaintimeperiodwithaprespecifiedprobability(e.g.,usuallydefinedas99.00%).Therefore,economiccapitalisoftencalculatedwithValueatRisk(VaR)typemodels.
2|P a g e
Capital modeling in banks surged as a necessity for the larger international financialinstitutions,whichdiscovered that theregulatoryapproaches takenbyregulatorswere toobasicandmainlynotriskbased.Forexample,creditriskcapitalrequirementsunderBaselIwerejustapercentage (8% times anothermultiplier)of the volumeof operations.Thismeasure,whichwasvery easy to calculate,wasnot risk sensitive, other than the differentiationof broad asset types.Therefore,complexbanksfoundthesecapitalrequirementstobeveryinefficientintermsofcapitalplanning, pricing, and leveraging limits and targets. With the evolution of the use of statisticalmodelsandavailabledata—especially inmarketriskmeasurement—regulatorsstartedacceptinginternal capital models developed by the big international financial institutions. Accordingly, in1996, an amendment was introduced to the Basel Accord (still Basel I) that allowed certainqualifying banks to calculate and hold capital in linewith their internalmodels. To differentiatethese measures of capital, banks started calling these internal calculations “economic capital,”because it had a very close relationship with the real economics of the business, whereas“regulatory capital” was the requirement mandated by regulators. As the business evolved, andregulations became more ample, complex financial institutions started relying more on theireconomic capital models for the measurement and management of risks, while simultaneouslyhavingtoholdregulatorycapital.Inmostcases,thedifferencesbetweenthesetwokindsofcapitalfor the same risk were very significant. This fact was one of the main motivators of Basel II,promptedmainlybyarequestfromthemorecomplexbanksthattheInternationalStandardsand,hence,bankingregulationsallowthemtousetheireconomiccapitalmodelstoallocateregulatorycapital. In otherwords, oneof theoutrightmotivations for theBasel II reformswas to close thepracticalgapbetweeneconomicandregulatorycapital.
As Basel II started to be implemented in most countries, the new regulatory paradigmestablished that banks—not just complex international financial institutions—must have IMMMprocessesforallmaterialrisks,andcalculateandallocateeconomiccapitalforeachandeveryoneoftheserisks.Foranygivenbank,theserisksaredefinedbyregulationsasidentifiedintheabove‐mentioned Core Principles: credit, market, operational, liquidity, interest rate, strategic,reputational, securitization, and so on. In this light, banks of any size, in virtually every country,needtoidentify,measure,monitor,andmitigatealltheserisks,andcalculate,evaluate,andallocateeconomiccapitalforeach.Thiscasediscussesasetofsimpleapproacheswithstraightforwardtoolsthat allow banks of any size and complexity to generate information for the management (theIMMMprocess)oftheserisks,andforthecalculationofeconomiccapitalbasedonavailablebalancesheetandregulatoryinformation.
In lightof these InternationalStandards,whicharenow formal regulations invirtuallyeverycountryintheworld,weutilizeaspectrumofbasicandmorecomplexapproachestogenerateaneconomiccapitalmodelcalculatedontheformallydefinedriskdriversineachcaseandprovidingforrisksensitivecapitalresultsforeachrelevantrisk.Additionally, foreachrisk,throughasetofbasic information, a set of key risk indicators is generated and combinedwith the capitalmodelresultstoproducerelevantriskreports.Sinceregulationsstillrequiremanyinstancesofregulatorycapital,suchcalculationisstillprovidedalongwithBaselStandardsasanotherusefuloutputofthedesigned tools. Finally, The Basel Committee differentiates credit, market, and operational risksfromtherest,definingthesethreeasthemostrelevantinanygivenfinancialinstitution.Accordingto the Three Pillar design of Basel II, these are known as Pillar I risks. Under Basel II and III,economicandregulatorycapitalcanbeunifiedforPillarIrisks.Inotherwords,forthesethreerisks(credit,marketandoperational),economiccapitalmodelsaregivenbytheBaselAccordasawaytogeneratesomestandardizationofmethodologiesandcomparisonamongbanksandcountries.
on the type of asset treated (100% for uncollateralized loans, 50% for mortgages, 20% forinterbank, etc.). This approach, however, does not differentiate by risk within each category. Inorder to create amore risk‐sensitive approach, Basel II incorporated themain logic of portfoliomodels, where capital is the amount required to cover unexpected losses. Unexpected losses, inturn,arecalculatedas theresidualgivenby thedifferencebetween themeanand theconfidenceintervalofalossdistributionfunction.
ProjectEconomicAnalysisToolonModelingBankingRisk
Figure 1 illustrates the PEAT utility’s ALM‐CMOLmodule for Credit Risk—Economic RegulatoryCapital(ERC)GlobalSettingstab.Thiscurrentanalysisisperformedoncreditissuessuchasloans,credit lines, anddebtat the commercial, retail, orpersonal levels.Toget startedwith theutility,existingfilescanbeopenedorsaved,oradefaultsamplemodelcanberetrievedfromthemenu.Thenumberofcategoriesofloansandcredittypescanbesetaswellastheloanorcreditcategorynames,aLossGivenDefault(LGD)valueinpercent,andtheBaselcredittype(residentialmortgages,revolvingcredit,othermiscellaneouscredit,orwholesalecorporateandsovereigndebt).EachcredittypehasitsrequiredBaselIIImodelthatispublicknowledge,andthesoftwareusestheprescribedmodelsperBaselregulations.Further,historicaldatacanbemanuallyenteredbytheuserintotheutilityorviaexistingdatabasesanddatafiles.Suchdatafilesmaybelargeand,hence,storedeitherinasinglefileormultipledatafileswhereeachfile’scontentscanbemappedtothelistofrequiredvariables (e.g., credit issue date, customer information, product type or segment, Central Bankratings, amountof thedebtor loan, interestpayment,principalpayment, lastpaymentdate, andotherancillaryinformationthebankorfinancialservicesfirmhasaccessto)fortheanalysis,andthe successfullymapped connections are displayed. Additional information such as the requiredVaRpercentiles,averagelifeofacommercial loan,andhistoricaldataperiodonwhichtorunthedatafilestoobtaintheProbabilityofDefault(PD)areentered.Next,theExposureatDefault(EAD)analysisperiodicity is selectedas is thedate typeand theCentralBank ratings.DifferentCentralBanksindifferentnationstendtohavesimilarcreditratingsbutthesoftwareallowsforflexibilityinchoosing therelevant ratingscheme(i.e.,Level1may indicateon‐timepaymentofanexistingloanwhereas Level 3may indicate a late payment of over 90 days and, therefore, constitutes adefault). All these inputs and settings can be saved either as stand‐alone settings and data orincluding the results. Userswould enter a unique name and notes and save the current settings(previously savedmodels and settings can be retrieved, edited, or deleted, a newmodel can becreated,oranexistingmodelcanbeduplicated).Thesavedmodelsarelistedandcanberearrangedaccordingtotheuser’spreference.
4|P a g e
FIGURE1 Creditrisksettings.
CreditEconomicandRegulatoryCapital
Figure 2 illustrates the PEAT utility’s ALM‐CMOLmodule for Credit Risk—Economic RegulatoryCapital’sResultstab.Theresultsareshowninthegridifdatafileswereloadedandpreprocessedand results were computed and presented here (the loading of data files was discussed inconnectionwithFigure1).However,ifdataaretobemanuallyentered(aspreviouslypresentedinFigure 1), then the grey areas in the data grid are available formanual user input, such as thenumberofclientsforaspecificcreditordebtcategory,thenumberofdefaultsforsaidcategorieshistoricallybyperiod,and theexposureatdefaultvalues (totalamountofdebt issuedwithin thetotalperiod).Onecanmanuallyinputthenumberofclientsandnumberofcreditandloandefaultswithinspecificannualtime‐periodbands.Theutilitycomputesthepercentageofdefaults(numberofcreditor loandefaultsdividedbynumberofclientswithinthespecifiedtimeperiods),andtheaveragepercentageofdefault is theproxyused for thePD. IfusershavespecificPDrates touse,theycansimplyenteranynumberofclientsandnumberofdefaultsaslongastheratioiswhattheuserwantsasthePDinput(e.g.,a1%PDmeansuserscanenter100clientsand1asthenumberofdefaults).TheLGDcanbeuserinputtedintheglobalsettingsasapercentage(LGDisdefinedasthepercentageoflossesofloansanddebtthatcannotberecoveredwhentheyareindefault).TheEADis the total loans amount within these time bands. These PD, LGD, and EAD values can also becomputedusingstructuralmodelsasisdiscussedlater.ExpectedLosses(EL)istheproductofPD×LGD×EAD.EconomicCapital(EC)isbasedonBaselIIandBaselIIIrequirementsandisamatterofpublicrecord.RiskWeightedAverage(RWA)isaregulatoryrequirementperBaselIIandBaselIIIsuchas12.5×EC.ThechangeinCapitalAdequacyRequirement( CAR@8%)issimplytheratiooftheECtoEADlessthe8%holdingrequirement.Inotherwords,theRegulatoryCapital(RC)is8%ofEAD.
5|P a g e
Theresultsobtainedbythemodelallowfortheconstructionofkeyriskindicators,comparingbasic regulatory capital requirements with these economic capital requirements. Additionally,whencoupledwith the internalorexternal ratingmodels (orcredit scores)aprofileofexpectedandunexpectedlossesforeachproductorassettypecanbeconstructed.ThisisalsothebasisfortheapplicationofRAROCindicators,andtheeffectiveallocationofeconomiccapital,inlinewiththeinternationalstandardsandlocalregulatoryrequirements.
FIGURE2 EconomicRegulatoryCapital(ERC).
MarketRisk
Formarket risk, as a Pillar I risk, the requirements are similar to those for economic regulatorycapital. The particularities of market riskmake it, possibly, the one that is easier tomodel andcalculate,andtheonethathashadmoretooldevelopmentsofar.Thisisexplainedbythefactthatthe main input for market risk measurement andmodeling is market prices of assets or, morepractically,theirvolatilities.Therefore,thereisgreatpublicavailabilityofdata,asopposedtotheother Pillar I risks that do not have daily prices publically available. As an example, there is nopublicpricingofaparticulargroupofretailloansissuedbyaprivatebank.Yet,bothmodelingtoolsformarketandcreditriskarebasedonthesameapproach:utilizingpaststylizeddata toprojectfuturebehaviorundercertainassumptionsandwithinaconfidenceinterval.Logicallythen,marketrisk has a great bundle of information available and the potential to better test and calibratemodels.Aspresented,marketriskmodelstakeonaValueatRisk(VAR)approach.
Figure3illustratesthePEATutility’sALM‐CMOLmoduleforMarketRiskwhereMarketDataisentered.Usersstartbyentering theglobalsettings, suchas thenumberof investmentassetsandcurrency assets the bank has in its portfolio, that require further analysis; the total number ofhistoricaldatathatwillbeusedforanalysis;andvariousVaRpercentilestorun(e.g.,99.00%and
6|P a g e
95.00%). Inaddition, thevolatilitymethodof choice (industrystandardvolatilityorRiskMetricsvolatility methods) and the date type (mm/dd/yyyy or dd/mm/yyyy) are entered. The amountinvested (balance) of each asset and currency is entered and the historical data can be entered,copyandpastedfromanotherdatasource,oruploadedtothedatagrid,andthesettingsaswellasthe historical data entered can be saved for future retrieval and further analysis in subsequentsubtabs.
FIGURE3 Marketriskdata.
Figure4illustratesthecomputedresultsfortheMarketVaR.BasedonthedataenteredintheinterfaceshownasFigure3,theresultsarecomputedandpresentedintwoseparategrids:theVaRresultsandassetpositionsanddetails.ThecomputationscanbetriggeredtobererunorUpdated,andtheresultscanbeexportedtoanExcelreporttemplateifrequired.Theresultscomputedinthefirst grid are based on user inputmarket data. For instance, theVaR calculations are simply theAssetPosition×DailyVolatility× InverseStandardNormalDistributionofVaRPercentile×SquareRootoftheHorizoninDays.Therefore,theGrossVaRissimplythesummationofallVaRvaluesforall assets and foreign exchange–denominated assets. In comparison, the Internal HistoricalSimulation VaR uses the same calculation based on historically simulated time‐series of assetvalues.Thehistoricallysimulatedtime‐seriesofassetvaluesisobtainedbytheAsset’sInvestment×Asset Pricet‐1 × Period‐Specific Relative Returns – Asset’s Current Position. The Asset’s CurrentPositionissimplytheInvestment×AssetPricet.Fromthissimulatedtimeseriesofassetflows,the(1–X%)percentileassetvalueistheVaRX%.Typically,X%is99.00%or95.00%andcanbechangedasrequiredbytheuserbasedontheregionalorcountry‐specificregulatoryagency’sstatutes.
7|P a g e
FIGURE4 MarketValueatRisk.
Manycountriesissueregulationsformarketriskmeasurementandcapitalallocation,wherebysome standardizedmodels are suggested or even imposed, in linewith theBasel Standards.WeanalyzesuchanexampleinFigure5,wheretheregulatorymodelcanbeobtainedbyutilizingtheparametersgivenbytheregulator(i.e.,volatilitiesandholdingperiodsforgivencommonassets).Thestructureofthetoolallowsforthecomparisonofregulatory,internal,andstressedscenarios,givingtheanalystsalargearrayofresultstobetterinterpretriskmeasurement,capitalallocation,andfutureprojections.
CentralBankMarketRisk
Figure 5 illustrates the Central Bank VaR method and results in computing VaR based on usersettings (e.g., theVaRpercentile, timehorizonof theholdingperiod indays,numberofassets toanalyze,andtheperiodof theanalysis)andtheassets’historicaldata.TheVaRcomputationsarebasedon thesameapproachaspreviouslydescribed,and the inputs, settings,andresults canbesavedforfutureretrieval.
8|P a g e
FIGURE5 MarketCentralBankVaR.
AssetLiabilityManagement
AswithanyotherBasel‐definedrisk,KRIsareconstructedbasedonthe inputsandresultsof themodelingtool,andcanbedulymonitoredandreported, in linewiththeIMMMprocess.Liquidityand interestrateriskareusuallymanaged together ina functioncalledALM,short forAssetandLiabilityManagement. These two risks are closely intertwined, since liquidity riskmonitors theavailabilityofliquidfundstoconfrontdisbursementrequirements(usuallyinthreetimehorizons:immediate and intraday, short‐term structure, and long‐term structure), while interest rate riskmeasurestheimpactofthedifferenceinmaturities,orduration,forassetsandliabilities.
Figure 6 illustrates the PEAT utility’s ALM‐CMOL module for Asset Liability Management—Interest Rate Risk’s Input Assumptions and general Settings tab. This segment represents theanalysisofAssetLiabilityManagement(ALM)computations.ALMisthepracticeofmanagingrisksthatariseduetomismatchesbetweenthematuritiesofassetsandliabilities.TheALMprocessisamixbetweenriskmanagementandstrategicplanningforabankorfinancialinstitution.Itisaboutofferingsolutionstomitigateorhedgetherisksarisingfromtheinteractionofassetsandliabilitiesaswellas thesuccess in theprocessofmaximizingassets tomeetcomplex liabilities such that itwill help increase profitability. The current tab starts by obtaining, as general inputs, the bank’sregulatorycapitalobtainedearlierfromthecreditriskmodels.Inaddition,thenumberoftradingdays in the calendar year of the analysis (e.g., typically between 250 and 253 days), the localcurrency’s name (e.g., U.S. Dollar or Argentinian Peso), the current period when the analysis isperformedandresultsreportedtotheregulatoryagencies(e.g.,January2015),thenumberofVaRpercentiles to run (e.g., 99.00%), number of scenarios to run and their respective basis pointsensitivities(e.g.,100,200,and300basispoints,whereevery100basispointsrepresent1%),and
The most straightforward way to present ALM structures for liquidity and interest‐rate riskmanagementisthroughtheutilizationofGapcharts.AGapchartissimplythelistingofallassetsandliabilitiesasaffectedbyinterestratemovementsorliquiditymovements,respectively,orderedontime‐definedbuckets(i.e.,days,weeks,months,oryears).Typically,forinterestrateriskthereare two main management approaches: a shorter‐term structure analysis based on a moreaccounting‐sideperspective,usuallyreferredtoastheNIM(NetInterestMargin)approach,andalonger‐termstructureanalysisbasedonamoreeconomic‐sideperspective,usuallyreferredtoastheEVE(EconomicValueofEquity)approach.TheNIMapproachrestsonthelogicthatthenaturalmismatchbetweenassetsandliabilitieshasanimpactonearnings,throughthenetinterestmargin,and such impact can be measured through given deltas (variations) in the referential marketinterestrate.Inthiscase,measuredthroughtheGAPchart,asappliedtobalancesheetitemsoftheassetandliabilitysidesrespectively.So,ontheonehand,anaturalNIMapproachwoulddeliverabalance sheet impact on earnings, based on the structure andmaturity of assets and liabilities,whensubjectedtoa100basispointincreaseinthereferentialmarketinterestraterisk.SincetheGapanalysisdefineswhich sideof thebalance sheet (assetsor liabilities)haspreponderance for
10|P a g e
eachtimebucket,analystscandefinewhichsignwouldapplytoearningsshouldinterestratesgoupordown.Therefore,thecombinationofthesetwotoolsallowsfortheestablishmentofdifferentbusinessandstressscenariosand,hence,thedeterminationoftargetsandlimitsonthestructureand duration of assets and liabilities. The EVE approach, on the other hand, is a long‐termevaluationtool,bywhichanalystscandeterminetheimpactoncapital(orequity,definedasassetsminus liabilities)of referentialmarket interest ratevaluations, as it affects thenetpresent valueanddurationofthedescribedbalancesheet items.Bythisapproach,thesystemcancalculatethedeltasindurationsandinnetpresentvalueofassets,liabilities,andequity,asmeasuredintheGapcharts.Therefore,suchvariationsallowfortheconstructionofscenariosforthedifferentimpactsonequityvalueanddurationofchanges in thereferentialmarket interestrate.TheseresultsarethenfedintodifferentKRIsformonitoring,defining,andcalibratingtargetsandlimits,inlinewiththeIMMMriskmanagementstructure.
Figure 7 illustrates the Gap Analysis results of Interest Rate Risk. The results are shown indifferentgrids foreach localcurrencyand foreigncurrency.GapAnalysis is,ofcourse,oneof themost common ways of measuring liquidity position and represents the foundation for scenarioanalysisandstress‐testing,whichwillbeexecutedinsubsequenttabs.TheGapAnalysisresultsarefrom user inputs in the input assumptions tab. The results are presented for the user again forvalidation and in amore user‐friendly tabular format. The Economic Value of Equity results arebased on interest‐rate risk computations in previous tabs. The impact on regulatory capital asdenotedbyVaRlevelsonlocalandforeigncurrenciesarecomputed,asarethedurationgapsandbasispointscenariosaffectingthecashflowsoflocalandforeigncurrencies.
Figure 8 illustrates theNet IncomeMargin (NIM) Input Assumptions requirements based oninterest‐rateriskanalysis.Thehighlightedcellsinthedatagridrepresentuserinputrequirementsfor computing the NIM model. The Economic Value of Equity and Gap Analysis calculations
11|P a g e
describedabove are for longer‐term interest‐rate risk analysis,whereas theNIMapproach is forshorter‐term(typically12months)analysisofliquidityandinterest‐rateriskeffectsonassetsandliabilities.
Figure 9 illustrates the PEAT utility’s ALM‐CMOL module for Asset Liability Management—LiquidityRiskInputAssumptionstabonthehistoricalmonthlybalancesof interest‐ratesensitiveassets and liabilities. The typical time horizon is monthly for one year (12 months) where thevariousassetssuchas liquidassets(e.g., cash),bonds,and loansare listed,aswellasotherassetreceivables. On the liabilities side, regular short‐term deposits and timed deposits are listed,separated by private versus public sectors, as well as other payable liabilities (e.g., interestpayments and operations). Adjustments can also be made to account for rounding issues andaccounting issues that may affect the asset and liability levels (e.g., contingency cash levels,overnightdeposits,etc.).Thedatagridcanbesetupwithsomebasicinputsaswellasthenumberofsubsegmentsorrowsforeachcategory.Asusual,theinputs,settings,andresultscanbesavedforfutureretrieval.
TheLiquidityRisk’sScenarioAnalysisandStressTestingsettingscanbesetuptotestinterest‐ratesensitiveassetsandliabilities.Thescenariostotestcanbeenteredasdataorpercentagechanges.Multiplescenarioscanbesavedforfutureretrievalandanalysisinsubsequenttabsaseachsavedmodelconstitutesastand‐alonescenariototest.Scenarioanalysistypicallytestsbothfluctuationsin assets and liabilities and their impacts on the portfolio’s ALMbalance,whereas stress testingtypically tests the fluctuations on liabilities (e.g., runs on banks, economic downturns wheredeposits are stressed to the lower limit) where the stressed limits can be entered as values orpercentagechange from thebasecase.Multiple stress tests canbesaved for futureretrieval andanalysisinsubsequenttabsaseachsavedmodelconstitutesastand‐alonestresstest.
Figure10illustratestheLiquidityRisk’sGapAnalysisresults.Thedatagridshowstheresultsbased on all the previously saved scenarios and stress test conditions. The Gap is, of course,calculatedas thedifferencebetweenMonthlyAssetsandLiabilities,accounting foranyContingencyCredit Lines. The gaps for the multitude of Scenarios and Stress Tests are reruns of the samecalculationbasedonvarioususerinputsonvaluesorpercentagechangesasdescribedpreviouslyintheScenarioAnalysisandStressTestingsections.
Figure11illustratestheAnalyticalModelstabwithinputassumptionsandresults.ThisanalyticalmodelssegmentisdividedintoStructural,Time‐Series,Portfolio,andAnalyticsmodels.ThecurrentfigureshowstheStructuralmodelstabwherethecomputedmodelspertaintocreditrisk–relatedmodelanalysis categoriessuchasPD,EAD,LGD,andVolatilitycalculations.Undereachcategory,specificmodelscanbeselectedtorun.Selectedmodelsarebrieflydescribedanduserscanselectthenumberofmodelrepetitionstorunandthedecimalprecisionlevelsoftheresults.Thedatagridin theComputations tab shows the area inwhichuserswould enter the relevant inputs into theselectedmodelandtheresultswouldbecomputed.Asusual,selectedmodels,inputs,andsettingscanbesavedforfutureretrievalandanalysis.
14|P a g e
FIGURE11 Structuralcreditriskmodels.
Figure11illustratestheStructuralAnalyticalModelstabwithvisualchartresults.Theresultscomputedaredisplayedasvariousvisualchartssuchasbarcharts, controlcharts,Paretocharts,and time‐series charts. Figure 12 illustrates the Time‐Series Analytical Models tab with inputassumptions and results. The analysis category and model type is first chosen where a shortdescriptionexplainswhattheselectedmodeldoes,anduserscanthenselectthenumberofmodelsto replicate aswell asdecimalprecision settings. Inputdata and assumptions are entered in thedatagridprovided(additionalinputscanalsobeenteredifrequired),andtheresultsarecomputedand shown. As usual, selectedmodels, inputs, and settings can be saved for future retrieval andanalysis. Figure 13 illustrates the Portfolio Analytical Models tab with input assumptions andresults. The analysis category andmodel type is first chosenwhere a short description explainswhattheselectedmodeldoes,anduserscanthenselectthenumberofmodelstoreplicateaswellas decimal precision settings. Input data and assumptions are entered in the data grid provided(additionalinputssuchasacorrelationmatrixcanalsobeenteredifrequired),andtheresultsarecomputedandshown.
Additionalmodelsare available in theCreditModels tabwith input assumptionsand results.Theanalysiscategoryandmodeltypearefirstchosenandinputdataandassumptionsareenteredintherequiredinputsarea(ifrequired,userscanLoadExampleinputsandusetheseasabasisforbuildingtheirmodels),andtheresultsarecomputedandshown.ScenariotablesandchartscanbecreatedbyenteringtheFrom,To,andStepSizeparameters,wherethecomputedscenarioswillbereturnedasadatagridandvisualchart.Asusual,selectedmodels,inputs,andsettingscanbesavedforfutureretrievalandanalysis.
15|P a g e
FIGURE12 Time‐seriescredit‐andmarket‐basedmodels.
FIGURE13 Creditportfoliomodels.
16|P a g e
OperationalRisk
Thecaseofoperationalriskisundoubtedlythemostdifficulttomeasureandmodel.Theoppositeofmarketrisk,by itsdefinition,operationalriskdata isnotonlyscarce,butbiased,unstable,anduncheckedinthesensethatthemostrelevantoperationalriskeventsdonotcomeidentifiedinthebalancesheetofany financial institution.Since themodelingapproach is still theVAR logic type,whereby themodelutilizespast empiricaldata toproject expected results,modelingoperationalriskisaverychallengingtask.Asstated,marketriskoffersdaily,publicauditedinformationtobemodeled. Conversely, operational risk events are, inmost cases, not public, not identified in thegeneralledger,and,inmanyinstances,notidentifiedatall.Buttheutmostdifficultycomesfromtheproper definition of operational risk. Even if we managed to go about the impossible task ofidentifyingeachandeveryoperational riskeventof thepast fiveyears,wewould stillhaveveryincomplete information. The definition of operational risk entails events generated by failure inpeople,processes,systems,andexternalevents.Withmarketrisk,assetspricescaneithergoupordown,orstayunchanged.Withoperationalrisk,anunknowneventthathasneveroccurredbeforecan take place in the studywindowandmaterially affect operations evenwithout it being a tailevent. So the logic of utilizing similar approaches for such different information availability andbehavior requires very careful definitions and assumptions. With this logic in mind, the BaselCommitteehasdefinedthat inorder tomodeloperationalriskproperly,banksneedtohave foursourcesofoperationalriskdata:internallosses,externallosses,businessenvironmentandinternalcontrol factors,andstressedscenarios.Theseareknownasthefourelementsofoperationalrisk,andtheBaselCommitteerecommendsthattheyaretakenintoaccountwhenmodeling.Forsmallerbanks, and smaller countries, this recommendation poses a definitive challenge, because manytimestheseelementsarenotdevelopedenough,ornotpresentatall.Inthislight,mostbankshaveresorted to just using internal data to model operational risk. This approach comes with someshortcomingsandmoreassumptions,andshouldbetakenasaninitialstepthatconsidersthelaterdevelopment of the other elements as they become available. The example shown in Figure 14looks at themodeling of internal losses as a simplified approach usually undertaken by smallerinstitutions. Sinceoperational risk information is scarceandbiased, it isnecessary to “complete”thelossdistributionswithrandomlygenerateddata.ThemostcommonapproachforthetaskistheuseofMonteCarlosimulationsthatallowfortheinclusionofmorestabledataandforthefittingofthedistributionsintopredefineddensityfunctions.
Figure14illustratestheOperationalRiskLossDistributionsubtab.UsersstartattheLossDatatabwherehistoricallossdatacanbeenteredorpastedintothedatagrid.Variablesincludelossesinthepastpertainingtooperationalrisks,segmentationbydivisionsanddepartments,businesslines,datesof losses, riskcategories, andsoon.Users thenactivate the controls to selecthow the lossdatavariablesaretobesegmented(e.g.,byriskcategoriesandrisktypesandbusinesslines),thenumber of simulation trials to run, and seed values to apply in the simulation if required, all byselectingtherelevantvariablecolumns.Thedistributional fittingroutinescanalsobeselectedasrequired. Then the analysis can be run and distributions fitted to the data. As usual, themodelsettingsanddatacanbesavedforfutureretrieval.
17|P a g e
FIGURE14 OperationalRiskdata.
Figure 15 illustrates the Operational Risk—Fitted Loss Distribution subtab. Users start byselectingthefittingsegmentsforsettingthevariousriskcategoryandbusinesslinesegments,and,based on the selected segment, the fitted distributions and their p‐values are listed and rankedaccordingtothehighestp‐valuetothelowestp‐value,indicatingthebesttotheworststatisticalfittothevariousprobabilitydistributions.Theempiricaldataandfittedtheoreticaldistributionsareshown graphically, and the statistical moments are shown for the actual data versus thetheoreticallyfitteddistribution’smoments.Afterdecidingonwhichdistributionstouse,userscanthenrunthesimulations.
Figure 16 illustrates the Operational Risk—Simulated Losses subtab where, depending onwhich risk segment and business line was selected, the relevant probability distribution resultsfromtheMonteCarlorisksimulationsaredisplayed,includingthesimulatedresultsonFrequency,Severity, and the multiplication between frequency and severity, termed Expected LossDistribution,aswellastheExtremeValueDistributionofLosses(thisiswheretheextremelossesin the data set are fitted to the extreme value distributions—see the case study for details onextremevaluedistributionsandtheirmathematicalmodels).Eachofthedistributionalchartshasitsownconfidenceandpercentile inputswhereuserscanselectone‐tail (right‐tailor left‐tail)ortwo‐tail confidence intervals and enter thepercentiles to obtain the confidence values (e.g., usercanenterright‐tail99.90%percentiletoreceivetheVaRconfidencevalueoftheworst‐caselossesonthelefttail’s0.10%).
Thesesimplemodelingtoolsallowsmallerbankstohaveafirstapproachatmoreadvancedoperationalriskmanagementtechniques.Theuseofinternalmodelsallowsforabettercalibrationof regulatory capital that knowingly overestimated for operational risk. The use of differentscenariosprovidingvariousresultscanallowsmallerbankstohaveamuchmoreefficientcapitalallocation for operational risk that, being a Pillar I risk, tends to be quite expensive in terms ofcapital,andquitedangerousatthesametimeifcapitalwasseverelyunderestimated.Togetherwiththe traditional operational riskmanagement tools, such as self‐assessment andKRIs, these basicmodelsallowforaproperIMMMriskmanagementstructure,alignedwiththelatestinternationalstandards.
![Validation of Internal Rating and Scoring Models ... of Internal Rating and Scoring ... [Basel II, §500] ... Basel I Advanced IRB Credit Risk AMA Operational Risk](https://static.documents.pub/doc/80x56/5ab075ad7f8b9a6b468b4b66/validation-of-internal-rating-and-scoring-models-of-internal-rating-and-scoring.jpg)
