Bogie Ozdemir, Senior Director

1
The Role of External Ratings Under Basel II

• Bogie Ozdemir, Senior Director
• Risk Solutions
• WDC
• Feb, 2008

2
Agenda

• External Ratings can play important roles in IRRs
  design, quantification and validation but we
  need to be careful about the appropriate usage.
• External Ratings can be used in
• Risk Rating Assignment
• Utilizing the Ratings
• Utilizing the Methodology
• Risk Rating Quantification
• Validation
• Benchmarking
• Becktesting

3
Use of the External Ratings in Risk Rating
Assignment
Example Risk Rating and PD assessment process

• Agency Ratings are used as inputs to the Risk
  Rating Assignment process along with PDs/EDFs
  from quantitative models, Financial rations and
  Qualitative Factors
• Pro All available information is utilized BUT..
• Con Double counting Financial Ratios and
  Qualitative Factors are already utilized in
  External Ratings. Financial Ratios may also be
  utilized in quantitative PDs/EDFs.

4
Use of the External Ratings in Risk Rating
Assignment

• Con - Consistency Issues Philosophical
  Differences Quantitative PDs/EDFs are more PIT
  than External Ratings.
• We dont have External Ratings and Quantitative
  PDs/EDFs for all obligors. Based on where we are
  in the cycle and whether or not we have External
  Ratings and Quantitative PDs/EDFs, the ratings of
  similar obligors can be inconsistent
• Utilizing rating methodology instead of the
  ratings for LDPs?

Quantitative PD model More PIT

• PD assigned to the Obligor
• More PIT (frequent re-grading or otherwise)
• More TTC (no re-grading or otherwise based on
  systematic factors)

Determine Risk Rating
Agency Ratings More TTC
PD assigned to the Risk Rating Long-run Unconditio
nal
Risk Rating to (Risk Rating) PD Mapping
Financial Ratios Maybe PIT but lagging
Qualitative Factors Intended to be PIT but not
frequently updated
5
Use of the External Ratings in Risk Rating
Quantification

• We need to assign PDs, LGDs and EADs to IRRS
  based on historical experience
• Internal data is limited and may not be
  sufficiently stressed
• External data (e.g. default rate time series for
  external ratings) are much longer How can we
  robustly utilize external data?

• The mappings between internal and external
  ratings are not always robust (usually
  methodologies are not mapped)
• Advantage of using rating agency methodologies
  for LDPs Internal ratings are calibrated to
  external data
• Miu and Ozdemir (2007)s approach enables
  estimations of Long-run PDs jointly from internal
  and external data

• Estimating and Validating Long-Run Probability
  of Default with respect to Basel II Requirements

6
Use of the Independent Models in Validation

• Benchmarking is the examination of the
  performance of risk rating systems relative to
  the comparable Risk Rating Systems and Models
• Benchmarking performing loans
• Benchmarking defaulted loans
• Backtesting is the examination of the performance
  of the Risk Rating System based on its historical
  data comparing realized and predicted outcomes.
• Benchmarking the backtesting performance.

7
Benchmarking Performing LoansUse of the External
Ratings

• We benchmark internal ratings against the
  external ratings when available and against SP
  Credit Estimates or CreditModel which produces
  Standard Poor's letter grade rating symbols, in
  lower case.

SP or CreditModel Ratings
Number of obligors
8
Benchmarking Performing loans, Monitoring

• An important aspect of benchmarking is to monitor
  the stability of the mapping between internal and
  external ratings over time. This helps to
  identify systematic trends and changes in trends.
  Several illustrative examples are provided below.

Trend 1 Increase in dispersion. The agreement
decreases.
Internal Risk Grades
External Risk Grades
Internal Risk Grades
Concentration of Obligors
External Risk Grades
Dispersion
Less Risky More Risky
Trend 3 Non - Parallel Shift. IRRS becomes less
conservative for low risk grades, and more
conservative for high risk grades.
Trend 2 Parallel Shift. IRRS becomes less
conservative.
Internal Risk Grades
Internal Risk Grades
External Risk Grades
External Risk Grades
9
Benchmarking Defaulted Loans

• Need to examine if realized default and migration
  rates ( realized LGD and EADs) are consistent
  with the relevant industry experience.
• We use models SP databases to provide this
  benchmark information
• CreditPro provides Default Rate, Rating
  Migration, and Default Correlation statistics
  customized by industry, geographic region, and
  time frame based on Standard Poors historical
  corporate rating and default data.

10
Use of the External Ratings in Validation
Backtesting

• Dimensions of validation, we need to test
• Discriminatory Power of the risk rating system,
• Calibration of the risk rating system,
• Risk Homogeneity of the obligors in each of the
  Risk Ratings and
• Realization of the Risk Rating Philosophy We
  measure our Risk Rating Philosophy relative to
  that of SP

• all information related to an obligors default
  likelihood is observable, and
• frictionless re-grading system

• no changes in ratings due to systematic reasons,
  (ie. Relative ranking does not change) but only
  for idiosyncratic (i.e. company-specific) reasons

11
Backtesting - Testing the Risk Rating Philosophy

• Mobility Metric Jafry and Schuermann (2003)
  demonstrate a method to re-scale the Euclidean
  distance to measure the mobility.
• The higher the concentration on the diagonal, the
  lower the mobility. A higher the mobility
  metric, the more PIT the risk rating system.
• By computing this single number, we can compare
  internal rating philosophy against the external
  benchmarks (SPs transition matrix). Example
  mobility metric for large corporate loan
  portfolios 0.318. where as mobility metric
  0.2 for the relevant the SP transition matrix.
  For suitable benchmarking we need to
• Common size the matrices compared - For the same
  range of PDs, a more granular RR system should
  show more migrations.
• Match the portfolio composition by industry when
  we construct the external TM. Can use CreditPro
  to obtain relevant benchmark transition
  matrices).

12
www.standardandpoors.com
Analytic services and products provided by
Standard Poors are the result of separate
activities designed to preserve the independence
and objectivity of each analytic process.
Standard Poors has established policies and
procedures to maintain the confidentiality of
non-public information received during each
analytic process.