portfolio loss distribution. risky assets in loan portfolio highly illiquid assets...
TRANSCRIPT
![Page 1: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/1.jpg)
Portfolio Loss Distribution
![Page 2: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/2.jpg)
Risky assets in loan portfolio• highly illiquid assets• “hold-to-maturity” in the bank’s balance sheet
OutstandingsThe portion of the bank asset that has already been extended to borrowers.
CommitmentA commitment is an amount the bank has committed to lend. Shouldthe borrower encounter financial difficulties, it would draw on thiscommitted line of credit.
![Page 3: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/3.jpg)
Adjusted exposure and expected loss
Let be the amount of drawn down or usage given default.
Asset value atlater time H, VH
Outstanding + commitment, Risky
(1) commitment, Riskless
Adjusted exposure is the risky part of VH.
Expected loss = adjusted exposure loss given default probability of default
* Normally, practitioners treat the uncertain draw-down rate as a known function of the obligor’s end-of-horizon credit class rating.
![Page 4: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/4.jpg)
Example calculation of expected loss
Commitment $10,000,000Outstanding $3,000,000Internal risk rating 3Maturity 1 yearType Non-securedUnused drawn-down on default (for internal rating = 3) 65%
Adjusted exposure on default $8,250,000EDF for internal rating = 3 0.15%Loss given default for non-secured asset 50%Expected loss $6,188
![Page 5: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/5.jpg)
Unexpected lossUnexpected loss is the estimated volatility of the potential loss in valueof the asset around its expected loss.
2EDF
22LGD LGDEDF AEUL
whereEDF).-(1EDF2
EDF
Assumptions* The random risk factors contributing to an obligor’s default (resulting
in EDF) are statistically independent of the severity of loss (as given byLGD).
* The default process is two-state event.
![Page 6: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/6.jpg)
Example on unexpected loss calculation
* The calculated unexpected loss is 2.16% of the adjusted exposure,while the expected loss is only 0.075%
Adjusted exposure $8,250,000
EDF 0.15%
EDF 3.87%
LGD 50%
LGD 25%
Unexpected loss $178,511
![Page 7: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/7.jpg)
Comparison between expected loss and unexpected loss
* The higher the recovery rate (lower LGD), the lower is the percentageloss for both EL and UL.
* EL increases linearly with decreasing credit quality (with increasing EDF)
* UL increases much faster than EL with increasing EDF.
Percentage loss per unit of adjusted loss
10%
5%
EL
UL
EDF10%
![Page 8: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/8.jpg)
Assets with varying terms of maturity
* The longer the term to maturity, the greater the variation in assetvalue due to changes in credit quality.
* The two-state default process paradigm inherently ignores the credit losses associated with defaults that occur beyond the analysis horizon.
* To mitigate some of the maturity effect, banks commonly adjusta risky asset’s internal credit class rating in accordance with itsterms to maturity.
![Page 9: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/9.jpg)
Portfolio expected lossii
ii
iip EDFLGDAEELEL
where ELp is the expected loss for the portfolio,AEi is the risky portion of the terminal value of the ith assetto which the bank is exposed in the event of default.
We may write
i i
ii
p
p
AE
ELw
AE
EL
where the weights refer to
.AE
AE
AE
AE
p
i
ii
iiw
![Page 10: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/10.jpg)
i A E i w i E L i E L i / A E i
1 $ 1 0 M 0 . 5 $ 1 0 . 12 $ 4 M 0 . 2 $ 0 . 5 0 . 1 2 53 $ 6 M 0 . 3 $ 0 . 6 0 . 1
M20$ iAE 1iw
i
ii
i
i
p
i
p
i
p
p
AE
ELw
AE
EL
AE
AE
AE
EL
AE
EL
105.01.03.0125.02.01.05.0 p
p
AE
EL
![Page 11: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/11.jpg)
Portfolio unexpected lossportfolio unexpected loss
i jjijiijp ww ULULUL
where2E
22LGD LEDFAEUL
ii DFiiii GD
and ij is the correlation of default between asset i and asset j. Due todiversification effect, we expect
.ULUL i
ip
![Page 12: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/12.jpg)
Risk contributionThe risk contribution of a risky asset i to the portfolio unexpected lossis defined to be the incremental risk that the exposure of a single assetcontributes to the portfolio’s total risk.
i
pii UL
ULULRC
and it can be shown that
.UL
ULUL
RCp
jijji
i
![Page 13: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/13.jpg)
Undiversifiable riskThe risk contribution is a measure of the undiversifiable risk of an asset in the portfolio – the amount of credit risk which cannot be diversified away by placing the asset in the portfolio.
i
ip RCUL
To incorporate industry correlation, using i industry and j industry
.U)1(UU
URC
kki
p
ii LL
L
L
![Page 14: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/14.jpg)
Calculation of EL, UL and RC for a two-asset portfolio
default correlation between the two exposures
ELp portfolio expected loss
ELp = EL1 + EL2
ULp portfolio unexpected loss
RC1 risk contribution from Exposure 1
RC2 risk contribution from Exposure 2
2122
21 ULUL2ULULUL p
pUL/)ULUL(ULRC 2111
pUL/)ULUL(ULRC 1222
ULp = RC1 + RC2
ULp << UL1 + UL2
![Page 15: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/15.jpg)
Fitting of loss distributionThe two statistical measures about the credit portfolio are
1. portfolio expected loss;
2. portfolio unexpected loss.
At the simplest level, the beta distribution may be chosen to fit theportfolio loss distribution.
Reservation A beta distribution with only two degrees of freedom isperhaps insufficient to give an adequate description ofthe tail events in the loss distribution.
![Page 16: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/16.jpg)
Beta distributionThe density function of a beta distribution is
otherwise0
0,0
10,)1(
),,(
11)()(
)(
xxx
xF
Mean and variance
.)1()( 2
2
1x
f(x, , )
![Page 17: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/17.jpg)
Economic CapitalIf XT is the random variable for loss and z is the percentage probability
(confidence level), what is the quantity v of minimum economic capital
EC needed to protect the bank from insolvency at the time horizon T
such that
Here, z is the desired debt rating of the bank, say, 99.97% for an AA
rating.
.]Pr[ zvXT
![Page 18: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/18.jpg)
XT
frequency of loss
ULp
ELp
EC
![Page 19: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/19.jpg)
Capital multiplierGiven a desired level of z, what is EC such that
.]ECELPr[ zX pT
Let CM (capital multiplier) be defined by
pULCMEC
then
.CMUL
ELPr z
X
p
pT
![Page 20: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/20.jpg)
Monte Carol simulation of loss distribution of a portfolio
1. Estimate default and losses
Assign risk ratings to loss facilities and determine their default probability
+ Assign LGD
and LGD
2. Estimate asset correlation between obligors
Determine pairwise asset correlation whenever possible
OR Assign obligors to industry groupings, then determine industry pair correlation
![Page 21: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/21.jpg)
3. Generate random loss given default
4. Generate correlated default events
Determine stochastic loss given default +
Correlated
default
events
+
Decomposition of covariance
matrix
+
Simulate default point
![Page 22: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/22.jpg)
5. Loss calculation
Calculate facility loss for each scenario and obtain portfolio loss
6. Loss distribution
Construct simulated portfolio loss distribution
![Page 23: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/23.jpg)
Generation of correlated default events
1. Generate a set of random numbers drawn from a standard normal distribution.
2. Perform a decomposition (Cholesky, SVD or eigenvalue) on the asset correlation matrix to transform the independent set of randomnumbers (stored in the vector ) into a set of correlated assetvalues (stored in the vector ). Here, the transformation matrix isM, where
The covariance matrix and M are related by
.MM T
ee
e e= M .
![Page 24: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/24.jpg)
Calculation of the default point
The default point threshold, DP, of the ith obligor can be defined asDP = N1(EDFi, 0, 1). The criterion of default for the ith obligor is
default if
no default if
ii DP'e
.'ii DPe
![Page 25: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/25.jpg)
Generate loss given defaultThe LGD is a stochastic variable with an unknown distribution.
A typical example may be
Recovery rate (%) LGD (%) LGD (%)
secured 65 35 21
unsecured 50 50 28
sisi f LGDLGDLGD
where fi is drawn from a uniform distribution whose range is selectedso that the resulting LGD has a standard deviation that is consistentwith historical observation.
![Page 26: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/26.jpg)
Calculation of loss
Summing all the simulated losses from one single scenario
LGDexposure Adjusted Loss
defaultin
Obligors
i
Simulated loss distribution
The simulated loss distribution is obtained by repeating the aboveprocess sufficiently number of times.
![Page 27: Portfolio Loss Distribution. Risky assets in loan portfolio highly illiquid assets “hold-to-maturity” in the bank’s balance sheet Outstandings The portion](https://reader035.vdocuments.net/reader035/viewer/2022062221/56649d9d5503460f94a86904/html5/thumbnails/27.jpg)
Features of portfolio risk• The variability of default risk within a portfolio is substantial.
• The correlation between default risks is generally low.
• The default risk itself is dynamic and subject to large fluctuations.
• Default risks can be effectively managed through diversification.
• Within a well-diversified portfolio, the loss behavior is characterized by lower than expected default credit losses for muchof the time, but very large losses which are incurred infrequently.