Title: City University of Hong Kong Professional Seminar on Latest Perspective on Basel II
1City University of Hong KongProfessional
Seminar onLatest Perspective on Basel II
- Simon Topping
- Hong Kong Monetary Authority
- 19 July 2004
2Outline
- Timetable for implementation in Hong Kong
- Range of approaches to be offered
- Transitioning from Basel I to Basel II
- Qualifying criteria for use of basic IRB
approaches - IRB validation qualitative quantitative
aspects / use of benchmarking - Determining minimum CARs under Basel II
- Changes to HKMAs supervisory approach
3Timetable for implementation in Hong Kong
- Implementation targeted for end-2006
- Following consultation with LegCo FA Panel,
preparing draft Banking Amendment Bill 2005
featuring rule-making power for HKMA in
relation to capital adequacy financial
disclosure - Consultation on implementation approach,
proposals for implementing IRB, draft Bill to
commence shortly - Basel II Consultation Group established
4Range of approaches to be offered
- Clearly desirable for all AIs to adopt Basel II,
but some smaller AIs have concerns about
cost/benefits - Default option of standardised approach for
credit risk ( operational risk, Pillar 2
Pillar 3) - AIs meeting qualifying criteria can apply for
approval to use either basic approach or IRB
approach - Therefore for credit risk there will be 4
options basic approach standardised foundation
IRB advanced IRB - While for operational risk there will be 2
options basic indicator approach standardised
approach (not AMA)
5Transitioning from Basel I to Basel II
- While the regulatory regime will be Basel
II-ready by end-2006, some AIs, particularly
those implementing the more advanced approaches,
will require an extended period to make the
necessary adjustments - Therefore proposing a 3-year implementation
period from end-2006 to end-2009 for IRB - During this period transitional arrangements
will apply - Generally speaking, all AIs will adopt
standardised approach at end-2006 unless they opt
for basic approach or indicate their intention to
adopt IRB - Pillars 2 3 will apply to all AIs from end-2006
6Qualifying criteria for basic approach
- Comprises Basel I treatment of credit market
risks plus operational risk plus Pillars 2 3 - Will be available to all AIs (primarily RLBs
DTCs, also some smaller banks) which are small
(total assets less than HK10bn) whose business
is simple - Not available for subsidiaries of larger banks
- Also available as an interim measure for AIs
planning to adopt IRB within the transitional
period
7Qualifying criteria for IRB
- Available to all AIs that can meet rigorous
qualitative quantitative qualifying criteria - AIs rating systems need to rank order quantify
risk in a consistent, reliable valid manner - Must provide for a meaningful differentiation of
borrower transaction characteristics, a
meaningful differentiation of credit risk,
reasonably accurate consistent quantitative
estimates of risk - Must have been in use for 2 years minimum of 2
years of data (within transitional period) - Must cover all material exposures (phased rollout
allowable, but IRB coverage must reach a target
level before transition to IRB allowed)
8IRB validation (qualitative aspect)
Qualitative Aspect
- Scope
- Coverage of asset classes
- Appropriate rating system design for AIs
exposures - Credible rating operations and process
- Adequate corporate governance and audit
- Adequate use of the rating system
- HKMAs validation methodologies
- Questionnaire for AIs self-assessment
- Checklist for on-site examination
9IRB validation (quantitative aspect)
Quantitative Aspect
- I. Data quality
- Data maintenance
- Use of external data
- sample data checking
- data storage process
- II. AIs internal stress tests used in assessment
of capital adequacy - Benchmarking against HKMAs internal
stress-testing parameters
III. AIs internal validation of PD/LGD estimates
internal statistical tests on discriminative
power of its credit scoring models
IV. HKMAs validation methodologies for PD/LGD
estimates
10IRB validation process
- A. HKMAs benchmarking models for identifying
underestimated PD/LGD - Listed companies (empirical testing a PD
term-structure model) - Private companies including SMEs (model based on
financial statements) - Retail exposures
- RML (empirical testing a model based on
expected-loss measures) - Credit cards, small SMEs, personal loans
(scoring systems) - Bank and sovereign exposures based on their
external credit ratings - Standard VaR validation for equities
IV. HKMAs validation metho-dologies for PD/LGD
estimates
- B. Benchmarking among AIs
- Comparing PD/LGD of same/similar exposures to
identify outlier with underestimated PD/LGD
measures - Results depend on individual AIs rating
approaches
- C. Back-testing
- Statistical tests (e.g. Gini coefficient)
- A sufficiently long period of actual default
history is necessary for meaningful tests
11Validating PD estimates (corporates)
- Among IRB AIs, we could compare PD of
same/similar exposures to identify outlier with
underestimated PD measures - However, it is possible that the use of
benchmarking among AIs may be constrained by the
widespread adoption of the same vendors, e.g.
Moodys KMV - AIs would not have enough actual default data for
meaningful back-testing during the initial period - We therefore need to develop some benchmarking
models which can be used to identify
underestimated PD measures (N.B. not developing
super credit risk models)
12Benchmarking of PD estimation (listed company)
Mapping with SPs default rates Map model PD
term structure of company to SPs default-rate
term structures of different ratings (static
pools cumulative average default rates)
Input market parameter Listed companys leverage
ratio its volatility
Implied 1-year benchmarking PD of company Based
on actual 1-year average default rate of assigned
rating
Assigning model SPs rating Based on mapping
result, a rating is assigned to the company
Model Engine Generate PD term structure of
company
Compare 1-year benchmark PD with AIs 1-year PD
of company based on its IRB system. Based on
comparisons for a number of companies, the
results will indicate any inconsistencies /
systematic underestimation in the AIs PD
estimates.
Empirical tests PD term-structure model based on
133 listed companies with credit ratings (BBB
below) in US with 1,337 data samples at different
time
13Determining minimum CARs under Basel II
- Charge for credit risk under standardised
approach is likely on average to be slightly
lower than under Basel I - However, this will be more than offset by the
charge for operational risk - Charge for credit risk under IRB less certain,
but unlikely to be significantly lower - Under Pillar 2, AIs will assess their target
overall level of CAR by means of a capital
adequacy assessment programme (CAAP) this will
be subject to supervisory review - Possible that less buffer or cushion above
the regulatory CAR will be maintained so CARs
may fall over time
14Changes to HKMAs supervisory approach
- For AIs on standardised approach, capital
adequacy requirements o/a credit risk will
continue to be set by the regulator - For AIs on IRB approach, however, cap ad
requirements o/a credit risk will be more
internally set (N.B. closer to economic capital) - For all AIs, setting the target CAR (under Pillar
2) will in the first instance be the
responsibility of the AI itself, rather than the
regulator although the regulator will conduct
its own assessment, at least initially (N.B. a
development of the risk-based approach) - Also for all AIs, market discipline (through
Pillar 3 disclosures) will play an increasing
role -