Credit Risk - Yats.com

❍Components of credit risk system and the evolution of credit risk measurement systems .... year and receives 120M DEM worth of Russian government.
145KB taille 166 téléchargements 458 vues
Financial Risk Management

Credit Risk

Following P. Jorion, Financial Risk Management Chapter 18

Daniel HERLEMONT

Introduction  Credit risk can be broadly defined as the risk of financial loss due to counterparty failure to perform obligations  Credit risk is far more important than market risk  Time and lack of diversification has been the primary reason for bank failure.  It is only recently that banking industry have learned to measure credit risk in the context of portfolio : that is VAR ...  Once measured credit risk can be managed and diversified like any other financial risk.  Banking sector is busily developing sophisticated internal models for credit risk Daniel HERLEMONT

Page 1 1

Challenges  Credit risk is much more difficult to quantify than market risk  There are many more factors driving risks,  some are extremely difficult to measure due to their infrequency.  credit risk model suffer from a verification problem. Unlike market risk for which backtesting can be performed on a daily basis, the longer horizon of credit risk model makes difficult to compare risk forecast with realization.  Nevertheless some progress have been accomplished  Credit risk is an active field of research Daniel HERLEMONT

Contents  Settlement risk during a short window  Components of credit risk system and the evolution of credit risk measurement systems  How to construct the distribution of credit loss for a portfolio  Correlations effect  How to manage credit risk  Basle II for credit risk Daniel HERLEMONT

Page 2 2

Settlement Risk  Pre-settlement risk  is the risk of loss due to the counterparty's failure to perform on an obligation during the lifetime of the transaction  default on loan or bond  failure to make the required payment on a derivative  can exist over a long periods, often years

 Settlement risk  is due to exchange of cash flows  is much shorter term  failure to perform on settlement can be caused by counterparty default, liquidity constraints or operational problems Daniel HERLEMONT

Settlement Risk Herstatt Bank failure 1974 March 1996, BIS report on settlement risk in FX market (>$1T/day), see www.bis.org/publ/cpss17.pdf Committee on Payment and Settlement Systems CLS, Target, netting, systemic risk. Real time gross settlement = RTGS Continuous linked settlement = CLS bank (1998) Daniel HERLEMONT

Page 3 3

Settlement Risk  Status of a trade  Revocable – can be canceled  Irrevocable - after the payment was sent but before the counter payment is due  Uncertain – after the payment from counterparty is due but before it is received  Settled – after the counterparty payment has been received.  Failed – after it has been established that the counterparty has not made the payment.

 Settlement Risk can occur during periods of irrevocable and uncertain status  Managed by  Real Time Gross Settlement (RTGS) systems  Netting Systems Daniel HERLEMONT

Settlement Risk - FRM questions

Daniel HERLEMONT

Page 4 4

Drivers of credit risk  Credit risk measurement systems attempt to quantify the risk of losses due to counter party default  The distribution of credit risk can be viewed as a compound process driven by:  Default which is a discrete state for the counterparty  Credit Exposure also known as Exposure At Default (EAD) which is the economic value of the claim on the counterparty at the time of default  Loss Given Default (LGD) which represent the fractional loss due to default  LGD = 1 - recovery rate

Daniel HERLEMONT

Credit Risk Measurement Tools - steps  Notional amount  say 8% of the notional amount was applied to establish the amount of required capital to face credit risk  problem: this approach ignores the probability of default

 Risk-weighted amounts  in 1998 the Basle Committee defined a rough credit risk classes and to apply different weights according to the classes  was too simple, creating an incentive to shift toward bad quality loans or bonds (e.g. there are no difference between AAA and C corporate credits)

 External/internal credit ratings  2001 Basle Committee proposed that bank uses their own Internal Rating Based (IRB) system  more in line to economic measures

 Internal portfolio credit models  However, the IRB approach is still a stand alone approach, withoutt considering correlation effects  This harks back to the age of finance before the benefits of diversification were formalized by Markowitz !!! Daniel HERLEMONT

Page 5 5

Credit Risk vs Market Risk

Daniel HERLEMONT

Credit vs Market returns

Frequency

Typical credit returns

Typical market returns

Source: CIBC

Portfolio Value

Credit return are highly skewed and fat tailed Daniel HERLEMONT

Page 6 6

Measuring Credit Risk - Credit Losses

N

CreditLoss = ∑ bi ⋅ CEi ⋅ (1 − f i ) i =1

bi is 1 if default occurs, 0 otherwise CE = credit exposure at the time of default f = recovery rate, (1-f) = LGD N

E [CL] = ∑ pi ⋅ CEi ⋅ (1 − f i ) i =1 Daniel HERLEMONT

Joint Events

Daniel HERLEMONT

Page 7 7

FRM-00, Question 46 An investor holds a portfolio of $50M. It consists of A-rated bonds ($20M) and BBB-rated bonds ($30M).Assume that the one-year probabilities of default are 2% and 4% respectively and are independent. The recovery rate for A-bond is 60% and recovery rate for BBB-bond is 40%. What is the one-year expected credit loss of this portfolio?

A. $672,000 B. $742,000 C. $880,000 D. $923,000 Daniel HERLEMONT

FRM-98, Question 42 A German Bank lends 100M DEM to a Russian bank for one year and receives 120M DEM worth of Russian government securities as collateral. Assuming that the 1-year 99% VaR on the Russian government securities is 20M DEM and the Russian bank’s 1-year probability of default is 5%, what is the German bank’s probability of losing money on thios trade over the next year?

A. Less than 0.05% B. Approximately 0.05% C. Between 0.05% and 5% D. Greater than 5% Daniel HERLEMONT

Page 8 8

Credit Risk Diversification Single loan of $100M, with probability of default 1% and 0 recovery. The expected loss and st. dev are: EL = 1%× ×$100M = $1M,

SD = $10M

Consider 10 loans, each for $10M. The total notional is $100M. Assume that defaults are independent with probability 1% and 0 recovery: EL = 10 ×1% ×10 = $1M,

SD = $3M

Daniel HERLEMONT

Page 9 9