Review Questions 31.docx

Review Questions 31.docx

《Review Questions 31.docx》由会员分享，可在线阅读，更多相关《Review Questions 31.docx（9页珍藏版）》请在冰豆网上搜索。

ReviewQuestions31

FIN2108Module3ReviewQuestions

ReviewQuestions3.1

3.1.1Pleaseattemptquestions1,2,3,5,8,10onpages106-7ofSathyeetal..

3.1.2Withregardtocreditscoring,distinguishbetweentype1andtype2errorsanddiscusshowtheyareaffectedbycreditpolicy.

3.1.3Withregardtobiasthatcanariseduetotheapplicationofcreditscoringmodels,whataresomegeneralsafeguardsthatlendinginstitutionscaninstigatetomitigatethisproblem?

3.1.4Listsomeofthecommoncategoriesofinformationusedtobuildacreditscore.

3.1.5Whichiseasiertoestimate,theprobabilityoftype1errorsortype2errors?

3.1.6

FICO（previouslyFairIsaacCorp）isapublicly-tradedcorporationthatcreatedthebest-knownandmostwidelyusedcreditscoremodelintheUnitedStates.

Howisthescoreconstructed,i.e.,whatarethemainconsiderations?

OutlinetherangeofFICOscoresandthekeyboundaryscores.

Whataresomestrategiesforimprovingacreditscoreunderthissystem?

3.1.7PleaseattemptProblemIICreditReportonpage693ofKochandMacdonald.

Creditscoringtechniques

ReviewQuestions3.1.1

Q1.Whatisstatisticalcreditscoring?

Howdoesitdifferfromjudgementalmethods?

A.Creditscoringisamathematically-basedtoolthatrankstheprobabilityofanunknownoutcome,e.g.aloanispaidordefaultedupon;thisisdonebyallocatingapointssystemtoknownvariables.Thecreditinformationofanapplicantisassessedandgradednumerically

togainatotalscore.Thisscoreisthenassessedagainsttheexpectationsoroutcomesofafinancialinstitutionbasedonitsriskappetite.

Statisticalcreditscoringisbasedontechnologyandinformationaveragesforprobabledefaultoutcomesoverarangeofloans.Judgementaldecision-makingreliesontheinputoftheindividualloanofficerbasedonrelationships,knowledgeoftheclientandtheexperienceoftheindividuallender.

Thescoregiventoacustomerisnormallyproportionaltothenaturallogarithmoftheodds（orlogit）ofthecustomerbecoming‘bad’.Itisnormallya3or4digitnumber;thehigherthenumber,thehigherthequalityoftheborrower

Q2.Doestheadoptionofcreditscoringaddvaluetoafinancialinstitution?

Whatpotentialexistsforanadverseoutcome?

Implementationofafullcreditscoringsystemacrossafinancialinstitutionwilladdvaluebecauseitwillallowfortheprobabilityofdefaulttobebuiltintothepricingoftheloan.

Thisallowsfordynamicprovisioningofloandefaulttobebuiltintothecashflowsoftheinstitution.Otherbenefitsareoutlinedinquestion3.

Anadverseoutcomecanoccuriftheassumptionsbuiltintothemodelareincorrectorbasedonoutofdateoranecdotalinformation.Thiscouldallowforanincreasedlevelofdefaultnotfactoredintothesystem.

Q3.Creditscoringmethodshavemushroomedinrecentyears.Whatarethreeapplicationsofthedifferingmethods?

Howdotheyaddvaluetothefinancialinstitution?

Applicationsinclude:

☐todeterminewhethercreditshouldbeapproved,

☐inthesettingofcreditlimitsoncreditcards/storecards,

☐behavioralmodelingsuchascollectionsscoring:

（whichinvolvespredictingacustomer'sresponsetodifferentstrategiesforcollectingowedmoney）,

☐inthepre-approvalofadditionalcredittoacustomer.

Theyaddvaluetothefinancialinstitutionbecause:

∙Creditscoringcanincreasereturnsbyidentifyinggoodversusbadpropositions.

∙Accuratecreditscoringreducescreditrisktoastatisticallyaccuratepercentageofoutstandingloans.

∙Creditscoringreducescost,savestimeandallowsforvolumeincreases.

∙Creditscoringincreasesflexibilityandmarketpenetration.Forexample,alowcostexpansionintothesmallbusinesssector.

∙Poorcreditoutcomesallowforcorrectionandre-adjustment（ofthemodel）bytheinstitution.

∙Strongerrelationshipscandevelopusingdifferentrelationshipcriteria.Loansstaffhavemoretimetodeveloprelationshipswithcustomers,andcanbetrainedinthisregard.

Q5.Whataretheadvantagesanddisadvantagesofimplementingcreditscoringwithinafinancialinstitution?

SomeadvantageswerementionedinQ3.Furthertothis:

A.Advantages:

∙Lesslabourcosts

∙Lowerlevelsofinvestmentintrainingoflendingofficers

∙Abilitytogrowtheloanportfolioswithlessresources

∙Loanpricingcanreflectmoreaccuratelytheriskinvolvedinextendingcredit

∙Abilitytoapplystandardisationacrosstheloanportfolio

∙Allowsuseofexperiencedstaffatthecentralisedlevel

∙Lossesaremanagedwithstatisticallyacceptableboundaries

∙Riskandreturncanbequantifiedacrosstheportfolios

∙Ameasurablerejectionmethod

Disadvantages:

∙Thepotentialforinappropriateanalysisanddecision-making–eg.ethnicbias

∙Potentialalienationofthetargetedclientgroupeg.smallbusinesses

∙Potentia