【正文】 hen ce, a di s t ri b u t i o n is nee d ed to repr es en t L G D . L ack of s en s i t i v i t y ana l y s i s w i t h resp ect to L G D . L ack of h i s t o ri cal dat a to val i d at e cu rren t l y us ed mod el s .Ri s k rati n g s , ex p ect ed defa u l t freq u en cy (E D F), an d mi g rat i o n p ro b ab i l i t i esIn det ermi n i n g E D F and mig rat i o n pro b ab i l i t i es , In t ern al rati n g sy s t ems may n o t be accu rat e or h av e en o u g h hi s t o ry . E D F and mig rat i o n pro b ab i l i t i es of p u b l i cl y trad ed bo n d s may n o t be accu rat e for ban k cred i t s . Mos t sy s t ems co mb i n e E D F and L G D . Mi g rat i o n and d efau l t co rrel at i o n sIs it reas o n ab l e to us e eq u i t y in fo rmat i o n to est i mat e co rrel at i o n s for b an k cred i t s ? L ack of h i s t o ri cal dat a to val i d at e mod el s us ed to est i mat e th i s p aramet er.Bank of Thailand 10 Risk Management Symposium September 2023 Current Issues in Credit Risk Modelling Adapted from “Credit Risk Modelling: Current Practices and Applications”, April 1999, by Basle Committee on Banking Supervision To pi c P a ra m eter Speci fi ca ti o n Is s ues / C o ncernsCred i t sp read s Fo r Mark t o Mark et mod el s , h o w much sp read sh o u l d be u s ed to val u e lo an s at each cred i t rati n g ? A re th e for w ard sp read s (bas ed on to d ay yi el d curv e) a g o o d app ro x i mat i o n of t h e fut u re sp read s ? H o w is l i q u i d i t y el emen t of cred i t sp read s tak en in t o acco u n t ?E x p o s u re le v el s D i fferen t in s t ru men t s (esp eci al l y marke t dri v en in s t ru men t s ) hav e di fferen t l ev el s of ri s k exp o s u re (. swaps vs lo an s ). E s t i mat es are mad e to make d i fferen t in s t ru men t s p arab l e. T h e accu racy of es t i mat es is qu es t i o n ab l e.Co mp u t at i o n al req u i remen tSo me mod el s are co mp u t at i o n al l y in t en s i v e.Bank of Thailand 11 Risk Management Symposium September 2023 Current Issues in Credit Risk Modelling Adapted from “Credit Risk Modelling: Current Practices and Applications”, April 1999, by Basle Committee on Banking Supervision To pi c V a l i da ti o n Is s ues / C o ncernsBack t es t i n g T o dat e, t h ere i s no w ay to veri fy accu racy . L i mi t ed ava i l ab i l i t y of h i s t o ri cal d at a is a b i g hu rd l e. G i v en a l i mi t ed hi s t o ry , t h e qu es t i o n is ho w to ade q u at el y b ack t es t .St res s tes t i n g St res s tes t i n g sh o u l d be u s ed at l eas t to part i al l y p en s at e for sh o rt co mi n g s i n ava i l ab l e ba ck t es t i n g meth o d s . Few in s t i t u t i o n s are d o i n g st res s tes t i n g s .Sen s i t i v i t y an al y s i sT h e ex t reme ta i l of t h e pro b ab i l i t y den s i t y fun ct i o n is li k el y to be h i g h l y s en s i t i v e to key ass u mp t i o n s and to est i mat es of k ey para met ers . Sens i t i v i t y an al y s i s is , t h erefo re, cru ci al in val i d at i n g a mo d el . V ery li mi t ed w o rk has b een p l et ed in th i s area to dat e.Bank of Thailand 12 Risk Management Symposium September 2023 Credit Risk Models (A) Purpose of a credit risk model ? Measuring economic risk caused by – Defaults – Downratings ? Identifying risk sources and their contributions ? Scenario analysis and Stress test ? Economic capital requirement and allocation ? Performance evaluation (. RAROC) Bank of Thailand 13 Risk Management Symposium September 2023 Credit Risk Models (B) Common Components 1. Model structure Transaction 1 Transaction 2 ... Transaction 1 Transaction 2 ... Counterparty A Counterparty B …… Portfolio of several counterparties and transactions Correlations Bank of Thailand 14 Risk Management Symposium September 2023 Credit Risk Models 2. Quantitative variables/parameters Default probability/intensity (PD, EDF) Loan equivalent exposure (LEE) Loss given default (LGD), Recovery rate (RR), Severity (SEV) Loss distribution Expected loss (EL) Unexpected loss (UL), Portfolio risk Economic capital (EC) Risk contributions (RC), Contributory economic capital (CEC) Bank of Thailand 15 Risk Management Symposium September 2023 Credit Risk Models (C) Model from Insurance (Credit Risk+) Only two states of the world are considered default and no default. Spread changes (both due to market movement and rating upgrades/downgrades) are considered part of market risk. Default probability is modeled as a continuous variable. Bank of Thailand 16 Risk Management Symposium September 2023 Credit Risk Models (C) Model from Insurance (Credit Risk+) There are 3 types of uncertainty: 1. Actual number of defaults given a mean default intensity 2. Mean default intensity (only in the new approach!) 3. Severity of loss Bank of Thailand 17 Risk Management Symposium September 2023 Credit Risk Models (C) Model from Insurance (Credit Risk+) The whole loan portfolio can be divided into classes, each of which consists of borrowers with similar default risk. Hence, a portfolio of loans to each class of borrowers