Basel ii-12-mar-07

Basel II : Risk Model Validation Quiet Revolution in the Supervisory World State Bank of Pakistan KARACHI, March 12 th , 2007 Christian Marlier Senior Director Head of CRO Executive Office

Basel II : Impact on Authorities Scope of Validation Supervisory & Regulatory Authorities : Impact & Roles Harmonisation Risk Model Basel II Compliance General principles Organisational Compliance Technical Compliance Implementation Compliance Supervisory Review : Points of attention Policy & Organisation Models & Process People IT & Data Conclusion Agenda

The higher the pressure put on trust The more important transparency and accountability become “ PROVE ME” “ TELL ME” “ TRUST ME” High Low Low High Basel I to Basel II TRUST TRANSPARENCY “ SHOW ME” From “Trust Me” to “ Prove & Evidence Me World”

BASEL II Impact on Authorities Training Capital Calculation Pillar I Validation Co-ordiantion of Information Planning & Co-ordination of Supervisory activities Coherency Between local accounting & tax framework Liquidity Spread Capital Strengthen soundness & stability of international banking system Promote Stronger Risk Mgt Practices Supervisors Consolidated / Home – Host Monetary Authority – Central Bank Minimum Required Capital Supervisory Review Market Discipline

Basel II impact on authorities Home & Foreign View Home State Bank Home Private Bank Foreign Private Bank Home Supervisor Mother Private Bank Validation Host / Home Host Supervisor Liquidity Spread Capital Monetary Authorities Central Bank Sovereign Ratings

Consolidating Supervisor / Home and Host Supervisor Supervision at the level of each legal entity but effective supervision = at group level (Basel 2 - risk management centralised) Consolidated supervisor - responsibility at 2 levels: stand alone and consolidated (extra burden for the financial institution) Supervisors should act in a coherent way with harmonised regulation but : Strengthening the role of the consolidated supervisor = political impact & Host supervisors have (legally) different responsibilities: National supervisor reports to national political bodies, must act in the interest of the country Financial responsibility towards their government (guarantees, etc.) In case of crisis, the national supervisor will try to limit the impact on his country Countries where foreign financial institutions play a major role are the most sensitive to this call for harmonisation

CEBS - guidelines for cooperation between consolidating and host supervisors Consolidating supervisor conducts SREP (Supervisory Review and Evaluation Process) for the group as a whole Host supervisors conducts this review at the subsidiary level: perform the SREP at local level, assess the internal governance at local level, within the framework of the group review and challenge the bank’s ICAAP (adequate capital), taking into account the group’s level review the local plan to close the gaps and communicate its conclusions to the consolidating supervisor Home / Consolidating supervisor approve Application File after intensive review and dialogue and taking into account the feedback from host supervisors. Consolidating Supervisor / Home and Host Supervisor

Role of Home and Host Supervisor Dialogue “Square Box” Consolidating Supervisor Mother Company Supervisory Cooperation Act as one company Bank’s Subsidiary Host Supervisor College of supervisors Local solvency reporting CRD local implementation Current legal requirements Consolidated Solvency Reporting Supervisory review Single entry point Validation & permission

Basel II : Impact on Authorities Scope of Validation Supervisory & Regulatory Authorities : Impact & Roles Harmonisation Risk Model Basel II Compliance General principles Organisational Compliance Technical Compliance Implementation Compliance Supervisory Review : Points of attention Policy & Organisation Models & Process People IT & Data Conclusion Industry View on “ Supervisors” Expectations Agenda

Risk Model Basel II Compliance First Pillar Art. 349 to 500: Credit Risk - Internal Ratings-Based Approach Art. 620 to 639: Operational Risk General Principles Second Pillar Art. 396 to 399 : Stress Tests Art. 416 – 418 : Definition of default Compliance with Minimum Criteria Ownership of senior management in the structures Independence of risk modeller Integrity of the risk / rating process Benchmarking Comparison to best practices Risk measures Pricing Process Backtesting Ex- Ante & Ex- Post Model inputs & Model Result Stress Testing Methodology Review Model Appropriateness for business use Robustness Transparency & Complexity Consistency with existing models Organisational Who ? Implementation How ? Technical What ? Basel II Risk Model - Compliance / Validation

1 2 4 3 5 6 7 8 Business Lines & CRO Needs Risk Model Basel II Validation Multiple Stakeholders Organisational Compliance Risk Analythics Business Lines Central Risk Management Risk Committees External Third Parties Internal & External Audit Supervisory Authorithies Basel II

A ssessing the predictive ability & use historical experience , forward looking , discriminating power , reassessment when divergence from expected results , P rimary responsibility = the bank NOT supervisors I terative process changing market and operating conditions on-going process N o single method statistical tools other methods (back testing, benchmarking, … internalisation of ECAI’s ) combination of methods Q ualitative and quantitative not only a mathematical exercise must cover structures, procedures, controls, ... S ubject to independent review validation ≠ audit Risk Model Basel II Validation AIG subgroup : 6 princip es Technical Compliance

Types of models Target Scorecards (Model for Credit Application) Retail Pools model Retail Behaviour models Retail / Micro & Small SME Econometric / Statistical models Medium Enterprises Generic / Market Models Public Corporate Look a like Models Sovereign / Financials Expert / Holistic Models Corporate Simulation models Project Finance Manual Models Start Up Basel Probability of default Models (PD) Exposure at Default (EAD) Models Loss given Default (LGD) Models Build – Internal Models Re-use – Out of The Shelf Models Buy – Out of the shelf Models DATA Risk Model Basel II Validation Technical Compliance

Create the development sample Choose the rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance

Create the development sample Choose the rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance Important to maintain coherency between field and modeller : art. 379, 386 - 391 Rating Anomaly Managers Support the analysis of differences between model rating and experts opinion Technical issues (balance-sheet / nace /...) Logical issues (financial analysis) Undisclosed Credit Event Organise, support and tests of new models Support and Validate - override

Create the development sample Choose the rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance Definition of Default Basel II default Recovery default Timeline Back to non- default level Risk class End-of- year Timeline Performing loans Pass Special Mention Uncertain Doubtful

Create the development sample Choose the rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance Definition of Default Transitions from Recovery level to the the Basel II level

Create the development sample Choose the rating criteria (variables) Generic Model Build or Buy ? PD calibration Model validation How was it developed? How applicable is it to: a) The specific product? c) The local market? How can this be proven? What is the quality of documentation? “ Generic” models built on different product data “ Generic” models built on different country data New credit bureau “generic models” How to adapt a generic to local situation ? Risk Model Basel II Validation Technical Compliance

Model power measures Model integrity Validation samples Validation over time Very useful – a generally accepted measure If a validation sample is used, the Gini should be based on this – not the development sample No absolute rules for “satisfactory” model strength – but < 25% is poor, > 75% is good! Can be influenced by the default definition Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance ROC or CAP statistics, or Gini coefficients. B A Random Model Developed Model Perfect Model All Clients Defaults Powerstat (P-Stat) = A A+B

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance Does the model rank order? Has the model used sensible characteristics? Has an appropriate model methodology been used? Has validation down to characteristic analysis been undertaken – e.g. marginal chi-squared analysis ? Has it been tested on different good / bad definitions? Has it subsequently been used in practice by the bank?

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance 115 files 20 Rating Anomaly Managers (Ram) C orrelation : 80 % with range 65 % and 95 % depending on Ram’s Commercial Loss : error Type II Financial Loss: error Type I

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance 0% 2% 4% 6% 8% 10% 12% 14% 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0% 2% 4% 6% 8% 10% 12% 14% frequency model % 'defaults' Masterscale Mapping

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance Two types: “ Hold out samples” – from same pool of exposures as the development sample – but a random 25%. All modelling performed on 75% Validation on 25% Validation of characteristic analysis Validation on PD calibration Validation on model power “ Out-of-time samples” – from different pool of exposures as the development sample

Model power measures Model integrity Validation samples Validation over time Create the development sample Rating criteria (variables) Model Development PD calibration Model validation Risk Model Basel II Validation Technical Compliance A model should be re-validated each and every month after implementation and reviewed yearly Similar characteristic analysis Similar score to PD calibration Similar model strength Soft of Hard review ?

Credit Limit Credit Process Pricing & Performance Market Risk Monitoring Key uses Risk Model Basel II Validation Implemantation Compliance Revenues : Customer Pricing & Performance Analysis Operational Risk Profile Risk Awareness Number of losses Average loss Year 1 Year 2 Year 3 Banking 3 Banking 2 Banking 4 Banking 1 Banking 5 750 MM 500 MM 0 AAA AA A BBB BB B CCC CC C Nominal Amount Theoretical Risk Curve Implemented Delegation Power Credit Delegation Power Market Risk

Stress tests in assessment of Risk model adequacy Identify possible events or future changes in economic conditions Regular perform Historical Scenario’s 1994 Bond Crash 1997 THB Crisis 2001 September 11 Attacks Hypothetical Scenario’s Confidence Crisis Real Estate Crisis Tail Risk stress test Body Risk stress test Two approaches for stress testing Event driven approach Portfolio driven approach Event Risk parameters Portfolio loss Portfolio loss Risk parameters Event Risk Model Basel II Validation Implemantation Compliance

Supervisory Review Points of attention Consultants Internalisation Risk Appetite Application file Communication & Guidance … Scarcity of Ressources Turn Over Management Knowledge Quant Syndrom ... Outliers Detection ECAI Internalisation Overruling process Involvement of local team Change culture /spreadsheet bank ? … Central Documentation Repository Collateral management Reconciliation of Data OR Tight schedule with limited tests… Policy & Organisation People Data & IT Models & Porcesses Basel II Compliance

Conclusion Management empowerment is crucial for validation Model validation not an exact science : a model might assess relative quality of the counterparty but it cannot capture all elements as it is based on portfolio analysis : this means : on average over time Expert judgement is of critical importance : for modelling and for communication Data issues center around quantity not quality Regional difference in culture and modelling e.g. equity model versus debt model Use test of critical importance Cherry Picking / Materiality Issues Supervisory Teams : Quant and Process Specialist Interviews on the Spot Benchmarking with internal models or industry sampling ( Securitisation ECAI experience look-a-like)

Christian Marlier Head of CRO Executive Office, Counsellor to Chief Risk Officer, Member of Central Risk Management Team Former Director of Corporate and Institutional Banking Business Support. Business Support ensures that counterparty risk assessment (reputation, credit, operational and strategic) are timely managed from the inception of a customer to its daily monitoring Former Director of Credit Risk Management (Analytics, Reporting, Policy & Portfolio Management), Fortis Representative for Working Group On Capital Adequacy for Basel II Implementation (IIF) Extensive Financial Markets Experience ( Option Chief Trader, Corporate Chief Sales, Global Head of Market Research, Fortis Representative for Market Risk IIF Task Force - Basel I Market Amendment) Degrees in Applied Mathematics and Business Administration from the Catholic University of Louvain (Belgium) and the University of Western Ontario (Canada). Registered EC Basel II Expert: Senior lecturer for EC sponsored programs : SME Financing (Europe -ESBG), Czez Basel II (Czech Republic – BBA – National Czech Bank), Basel II & Risk Management (ESBG – China Banking Regulatory Commission) Senior Lecturer FEBELFIN & International Risk Confererences Publications : Treasury Risk Management 1996; Accounting, financial and fiscal aspects of Derivatives, 1996 Personal : Fencing’instructor (32-2) 565.56.00 3, Montagne du Parc B-1000 Brussels, Belgium [email_address]

Basel ii-12-mar-07

More Related Content

What's hot (14)

Similar to Basel ii-12-mar-07 (20)

More from Rod Medallion (20)

Basel ii-12-mar-07

Editor's Notes