Professional Documents
Culture Documents
AGENDA
3: APPLICATIONS FOR CREDIT RISK MEASUREMENT TECHNIQUES
SECTION
Exposures need constant monitoring Credit portfolios must be actively managed Active portfolio management requires accurate measurement of
portfolio risk and return
The Opportunity
Active credit risk portfolio management can lead to substantial improvement in return per amount of risk & more efficient use of economic capital.
Default Probabilities
SECTION
What is Risk?
Risk is defined as volatility in earnings
RISK P&C Risk
CREDIT RISK MARKET RISK OPERATIONAL RISK MORTALITY RISK MORBIDITY RISK BUSINESS RISK
For Emerging Market exposures ING splits Credit Risk in Credit Risk and Transfer Risk
Expected Loss:
UNEXPECTED LOSS
Results in volatility of return over time
EL
UL
Time
Differentiated Capital
Loss Rate
EL UL
Time
=
/
RAROC
Expected Loss
Borrower related
EXPECTED DEFAULT FREQUENCY 0.10% (A-)
Facility Related
EXPOSURE AT DEFAULT $100 mm LOSS GIVEN DEFAULT 60%
Expected (Credit) Loss has 3 components which should be derived bottom-up from the risk characteristics of individual transactions Expected Default Frequency is the probability that the borrower will default
- Derived from the companies borrower risk rating - Depends on the term of the facility
Loss Given Default is the percentage of Exposure at Default that is expected to be lost in case of a default by the borrower
- Depends on the seniority, and the type, quantity and quality of the cover
Internal Rating 1 2 3 4 5 6 7
6
-> risk
PRINCIPAL LOSS
COST OF CARRY
ADMINISTRATIVE COST
Most of the time, the portfolio has smaller than the Expected Loss Sometimes, the portfolio has losses equivalent to the Expected Loss
Probability
Year
$0
EL
Loss
$0
Loss
Facility UL
Portfolio UL
Facility ULC
Portfolio EC
Facility EC
Default Correlations
Unexpected Loss is defined as the standard deviation of actual loss and depends on the same variables as Expected Loss At transaction level it is calculated as follows:
UL = EAD EDF LGD LGD 2 0.25 + LGD 2 * ( EDF EDF 2 )
$2,049,302 = $100 mm * SQRT { 0.10% * (60% - 36%) * 0.25 + 36% * (0.10% - 0.0001%) }
AA-rating
EL
=> 0.05% default probability => 99.95% confidence level => capital multiple 7
=
=
Unexpected Loss
$2,049,302
x
*
Correlation x Factor
0.15% *
Capital Multiple
7
InterPortfolio Correlation
whereby correlation factor is based on insights gained from credit risk portfolio modeling using regression analysis; correlation factor is a function of rating, LGD, R-squared, term, industry, country
SECTION
Consistent bottom-up credit risk measurement is the foundation for a series of value added initiatives
The goal of managing credit risk concentrations is not to reduce loan losses, but to reduce their volatility
Studying default correlations is of great importance!
frequency of occurence
loss rate
Summary
Credit Risk Models are used by ING to build a better understanding of the volatility of credit losses Within ING Credit Risk Modeling is an important building block of the internal Economic Capital framework ING believes that Credit Risk Modeling will allow us in the coming years to increase the risk adjusted performance on our investment portfolios
But also
ING understands that models are as good as the assumptions that need to be made to run the models ING is aware that credit risk models will complement but never replace common sense in managing a business
Paul Narayanan
Investment Actuary Symposium/AFIR Colloquium November 10, 2004
Structured finance
Tranche risk analysis Analyzing deep out of the money puts
Ratio
Bankrupt NonBankrupt Re t u r n o n Asse t s -0.0055 0.1117 Ea r n in g s St a b ilit y 1.6870 5.7840 Debt Service 0.9625 1.1620 Cumulative Profitability -0.0006 0.2935 Liquidity 1.5757 2.6040 Capitalization 0.4063 0.6210 Size 1.9854 2.2220
11
Ratio (Median) EBIT Interest Coverage (x) EBITDA Interest Coverage (x) Free Oper Casf Flow/Total Debt (%) Funds from Operations/Total Debt (%) Return on Capital (%) Operating Income/Sales (%) Long Term Debt/Capital (%) Total Debt/Capital (%) Number of Companies
BBB 3.7 5.3 8.5 30.8 13.6 15.4 42.5 48.2 218
Market Value
Increase in servicing requirement due to ballooning term loan # of SD from Mean Shape of probability density in region of distress
EDF
Now
1 Year
Time
Some Observations
Driven by historical data, limited forward looking input Do not incorporate many external factors that are not easily quantified, but are nevertheless important (e.g., management quality, corporate governance, lack of transparency in corp. structure)
Credit Exposure
For most cases, credit exposure is straightforward for bonds, it is the principal plus accrued interest Credit exposure in derivatives can fluctuate as a result of interest and currency rates Credit exposure may exist in insurance products either directly or indirectly
10
The exposure would typically start at zero, reach a maximum and then decline to zero at maturity
Derivatives Exposure
Exposure = Current MTM + Potential Future Exposure
Start with stochastic models for interest rates Impose a correlation structure on the key rates Use Monte Carlo simulation Apply collateral, netting arrangements, guarantees Downgrade triggers are tougher to model
11
LGD of holding company and subsidiary debt is a complex issue involving both corporate structure and capital structure.
12
13
2 YP 0.0 4.6 48.7 56.7 62.3 64.1 71.6 0.0 37.9 52.9 57.7 59.7 65.0 77.6
Bank Debt Senior Secured Notes Senior Subordinated Notes Subordinated Notes Junior Subordinated Notes
25 Percent 35 70 75 80
14
15
16
17
Concluding Remarks
Credit models have had the most impact through the internal rating systems Other models while important, do not yet play a central role Data lags behind theory
18