Internal Models Approach for Capital Charge for Market Risk B. Mahapatra Reserve Bank of India July 21, 2010.

Internal Models Approach for Capital Charge for Market Risk

B. MahapatraReserve Bank of India

July 21, 2010

Structure1. Introduction2. Broad Principles for RBI Approval 3. Qualitative Criteria for IMA Approval and

Corporate Governance of Risk4. Risk Measurement5. Minimum Quantitative Criteria6. Back Testing7. Stress Testing8. Validation of Models

1.1. Introduction - Time Schedule for Advanced Approaches –

Circular July 7, 2009Sl. No. Approach Earliest Date of

Application to RBILikely Date of Approval by RBI

a. IMA for Market Risk April 1, 2010 March 31, 2011

b. TSA for Operational Risk April 1, 2010 September 30, 2010

c. AMA for Operational Risk April 1, 2012 March 31, 2014

d. IRB for Credit Risk Foundation & Advanced)

April 1, 2012 March 31, 2014

1.2. Guidelines Issued by RBI

• Guidelines for IMA for Market Risk – issued on April 7, 2010 by RBI

• Banks’ boards to take a decision whether and when to adopt IMA

• Banks can continue with simpler methods for capital charge for credit and operational risks, while adopting IMA for capital charge for market risk

1.3. IMA for Market Risk - Guidelines

• Based on – BCBS 1996 Guidelines; – Basel II Framework; and – Updated with July 2009 enhancements/revisions

to Basel II Framework; and – Indian position

1.4. Scope of Application

• Application at solo and consolidated levels (other than insurance business)

• No off-set allowed among and between group entities

• Solo – on net basis, except where repatriation not possible

• To monitor market risk exposure of each entity separately

1.5. Types of Risks Covered

• Interest rate-related instruments and equities in trading book

• Exchange rate risk (including open position in gold) both in banking and trading book

• Risk relating to investments in MFs in trading book

1.5.1. Interest Rate Risk in Trading Book

• General market risk – general level of movement in prices in the market

• Specific risk – movement of price due to factors related to specific issuer– Default risk – potential of direct or indirect loss– Credit migration risk – due to down(up)gradation– Credit spread risk – variation due to change in

spread over risk free security– Incremental risk – risks which are not captured in

VaR based estimates of specific risk

1.5.2. Equity Price Risk in Trading Book

• General market risk• Specific risk

1.5.3. Exchange Rate Risk – Both in Banking and Trading Book

• Open positions, including in gold

1.5.4. Commodity Price Risk

• Not applicable in India

1.5.5. Trading Book Under IMA

• Only HFT • AFS – to continue under SMM, as markets are

illiquid and market prices may not be available (Indian position)

1.6. Combination of IMA and SMM

• If other group level entities are not ready, IMA for parent, and those entities which are ready, and SMM for others

• Insignificant positions, minor currencies, negligible business areas, etc. can be under SMM

• Expected to follow IMA to all market risk positions and entities in due course

• Present plan of action to RBI• Capital charge – summation of IMA and SMM,

where applied• No reversal allowed from IMA to SMM, unless

RBI stops it on account of the IMA not working properly

• Inform RBI any change in IMA as approved by RBI

1.7. Prudential Floors - Indian Stipulation

Years Year 1 Year 2 Year 3

Prudential floor as percentage of minimum capital requirement as per SMM

100 90 80

2.0.Broad Principles for RBI Approval

• Assessment of bank’s risk management system – conceptual soundness and implementation integrity

• Availability of skilled staff – trading, risk control, audit, back office, etc.

• Track record of model’s accuracy• Regular conduct of stress testing• Period of initial monitoring and live testing

3.0. Qualitative Criteria for IMA Approval – C. G. of Risk

• Board and senior management oversight of risk management

• Documentation of policies, procedures and model parameters – Risk Management Manual

• Maintenance of market risk model (MRM) dossier – records of details, changes/refinements

3.1. Contents of MRM Dossier• Full technical specifications of the model• RBI approval, subsequent changes, and

conditions of RBI• Complete detail of record of changes

3.2. Organisation of Market Risk Function

• Front office• Middle office/risk control unit• Model construction unit• Model validation unit – could be part of risk

control unit• Back office• Internal audit – required qualification, skills

and experience

3.3. Information Technology System

• Feeder system; risk aggregation system; time series data base, VaR model system; stress testing system; back testing system; etc. are appropriate; data quality; reconciliation and checks on completeness of capture

• Systems development, change control and documentation; security audit trails; systems availability and contingency procedures; network adequacy

• Operational statistics relating to the VaR model production process, including statistics relating to timeliness, number of re-runs required and reliability of data feeds

4.1. Risk Measurement – General Principles

• Internal risk measurement model closely integrated into day-to-day risk management process of the bank and in conjunction with internal trading and exposure limits. Its out put integral part of planning, monitoring and controlling bank’s market risk profile

• Robust model validation and internal approval process and back test and stress testing for market risk measurement

• VaR in INR – take appreciation / depreciation in security into VaR input but not for accounting – Indian position

• Market risk and exposure limit well understood by traders and senior management

• Board and senior management understand the basis, major assumptions, strengths and limitations of the model

• Risk measurement system should be scalable – volume increase, new valuation, new product, etc., particularly the IT capability

• Staff, vendor and consultants should have the requisite skills

4.2. Use of Vendor Models• Same validation standards as internally

developed model• Obtain and keep on record all mathematical and

statistical basis of the model and ensure that bank’s staff understand these

• Document the role of vendor model, bank’s understanding of it, its appropriateness, regularly review

• Proprietary vendor models not providing documentation – not allowed

• For specific risk use SMM for now

4.3. Criteria for FX (and Gold) Net Open Position Exposures

• For each currency separately• Capture risk factor for each currency

4.4. Criteria for Equity Exposure

• General market risk– Modelled with a single index or multi-factor

model (more complex)– Use market indices, sector indices, or sub-sector

indices, in conjunction with market beta/sector beta/sub-sector beta

– In each equity market where the bank has significant position (>1% of NW)

– Capture volatility and correlation effect– Depending on its position

• For specific risk - use SMM for now

4.5. Investment /Exposure to MFs

• Follow SMM for now

5.1. Minimum Quantitative Criteria for VaR Models

• Capital charge will be a function of– Normal VaR (for general market and specific risks)– Stressed VaR (for general market and specific

risks)– IRC (for interest rate specific-risk capital charge)

• General market risk and specific risk can be modelled together while calculating Normal or Stressed VaR or separately

• In combined measure, bank should isolate VaR for general risk and specific risk for back-testing and use in day-to-day risk management

• As modelling and isolating specific risk under VaR will take some time, banks to start with:– Normal VaR measure (for general market risk)– Stressed VaR measure (for general market risk)– Specific risk capital charge as per SMM (even

if its VaR model captures specific risk but not able to isolate it)

5.2. Parameters for VaR Models

• Banks can use any model, subject to:– VaR captured on daily basis– 99th percentile, one tail confidence interval is used– Minimum 10-day holding period – square root

rule can be used for linear positions – for non-linear position periodically justify reasonbaleness to RBI

– Minimum 1 year historical observation period or weighted average time not less than 6 months

– Bank may use alternate weighing schemes so long as it is conservative

– RBI may stipulate short holding period, say 3 months, if situation so warrants (regime shift due to extreme volatility, etc.)

– Update data sets no less than once in 3 months or flexible enough to allow more frequent changes

– No particular type of model is prescribed – variance-covariance, historical simulation or Monte Carlo simulation

– RBI may recognize empirical correlation among risk categories, if banks can demonstrate

– Must capture unique risks in options• Gamma risk• Full 10-day price shock• Vega risk – volatility of rates and prices of the

underlying

5.3. Parameters for Computing Stressed VaR

• 10-day, 99th percentile, one tailed confidence interval VaR measure of the current portfolio

• Calculated at least weekly• Model input calibrated to historical data from

a continuous 12-month period of significant financial stress relevant to a bank’s portfolio

• Stress period requires RBI approval• Choice of VaR model left to banks

5.4. Calculation of VaR, Stressed VaR and Capital Requirement

• Normal VaR – higher of– Previous day’s VaR (VaRt-1) and– An average of the daily VaR of preceding 60 business

days (VaRavg) * multiplication factor (mc+pc)

• Stressed VaR – higher of– Latest VaR calculated (weekly) (sVaRt-1)– An average sVaR calculated over preceding 60

business days (sVaRavg) * multiplication factor (ms+ps)

• Capital requirement• C = max{VaRt-1;(mc+pc)*VaRavg} +

max{sVaRt-1;(ms+ps)*sVaRavg} where• mc and ms = multiplication factors set by RBI

depending on quality of model, with minimum 3

• pc and ps = plus or add on factor, ranging from 0 to 1, decided by bank depending on back-testing of Normal VaR model (not stressed VaR)

6.0. Back-testing under IMA

• Comparison of bank’s daily VaR measure with subsequent daily profit or loss (trading outcome) – the number times risk measures were larger than trading outcomes

Zone Number of exceptions Increase in multiplication factor over 3

Green 0 0.00

1 0.00

2 0.00

3 0.00

4 0.00

Yellow 5 0.40

6 0.50

7 0.65

8 0.75

9 0.85

Red 10 1.00

• If falls under red zone, RBI may require the bank to discontinue the model or begin work on improving the model

• RBI may require bank to increase capital if the model is not capturing risk properly

• No exceptions means the model is conservative or something wrong?

• Explanation for back-testing results• Quarterly reporting of back-testing results to

Board• Quarterly reporting of back-testing results to

RBI

7.0. Stress Testing for IMA

• General stress testing guidelines to be followed, in addition to those for IMA

• Visualise low probability events, certain stress events, extreme but plausible events

• Both quantitative and qualitative (whether capital can withstand the stress)

• Communicated to senior management and periodically to Board

• Provide information to RBI on general and specific scenarios

• General scenario– Need not confine to stress scenarios in India only– Historical range of volatilities and correlations

– Extreme values based on hostorical volatilities and correlations

• Specific scenarios– Scenarios most adverse to its portfolio– Should be reviewed periodically and reflected in

the policies and limits set by management and Board

8.1. Internal Validation of Models

• By qualified parties independent of model development to ensure conceptual soundness and adequacy – initially and periodically

• At a minimum– Review logical and conceptual soundness– Compare the model with another model or

benchmark model– Review the back-testing done on the model

• Responsibility of model validation unit– Ensuring that current systems set up is capable– All changes to models or modelling process should

be validated and approved– Maintain previous version of model being altered– Model subject to change control procedure –

computer codes can only be changed by authorised staff

8.2. External Validation of Models

• At a minimum– Verify that internal validation is satisfactory– Formulae used in calculation and pricing are

validated by Risk Control Unit, independent of traders

– Structure of model is adequate for bank’s activities and geographical coverage

– Compare back-testing results– Data flows are transparent and accessible to all

auditors

8.3. Documents for Application to RBI

• Preliminary and detailed application to RBI for prior approval

• Request letter• Internal audit report of the model• A MR File• MR Model Dossier

8.4. Model Validation by RBI• First give RBI notice of intention• External auditor’s report• RBI assessment– Assessment and accuracy of documentation– Assessment of model scope– Qualitative review– Assessment of technological environment and

information integrity– Quantitative review– Model monitoring – any modification proposed

need prior intimation and approval from RBI

8.5. Additional Pillar II Requirements

• Policies and procedures for trading book eligibility

• Valuation• Stress testing under IMA – RBI to examine

under SREP whether bank has enough capital for the IMA and take appropriate measure

8.6. Additional Disclosure under Pillar III

• Qualitative – portfolio covered, characteristics of models used, stress testing, back testing, scope of acceptance by RBI, etc.

• Quantitative – high, mean,and low VaR and stressed VaR over the reporting period and period-end, comparison with actual gains/losses experienced by the bank, out-liers in back-testing, etc.

8.7. Capital Allocation for Market Risk

• First allocate capital for credit and operational risk and then surplus should be sufficient for market risk

• Thank you