Ch&cie model pricing validation 20140922_risk & finance

18
Pricing model validation and process review 22th, September 2014 Stephane Eyraud [email protected] Benoit Genest [email protected]

description

In a complex control framework for pricing model validation, CH&Cie has adopted a dedicated approach meeting both business and regulatory expectations

Transcript of Ch&cie model pricing validation 20140922_risk & finance

Page 1: Ch&cie model pricing validation 20140922_risk & finance

Pricing model validation and process review

22th, September 2014

Stephane Eyraud [email protected]

Benoit Genest [email protected]

Page 2: Ch&cie model pricing validation 20140922_risk & finance

2

Pricing & Validation key issues

Approach

Appendixes

CH&Cie presentation: risk management offer

Agenda

1

2

3

4

Page 3: Ch&cie model pricing validation 20140922_risk & finance

3

Finance delegates to CIB Finance the supervision of the entire P&L and valuation control process CIB Finance responsibility is carried out through “CIB Financial Control” and the coordination of the

governance structure CIB Financial control is a “global” finance control function which is responsible for supervision of the

entire Valuation and P&L Control framework (which includes 1st and 2nd level controls) across capital market activities, global coordination (prepares and drives the monthly Committees that examine all issues relating to valuation, P&L and system booking)

The responsibility is shared by many players, each of them is responsible for their respective perimeter

Operations

Front Office

Risk

Finance

Middle Office & Product control

Back Office

Global FinanceHeadquarter

Global Finance Control

CIB Local Finance

Global Finance Control

Local

1

2

3

4Based on the charter of responsibilities, which defines the breakdown of responsibilities on the valuation and P&L controls, the organization is placed under the supervision of the Finance function Finance guarantees the

production and the quality of the Group financial statements and Group Management accounts

Finance uses to delegate the production and control of the financial instruments’ fair value, to the various participants

Finance delegates to Risk the authority to control the fair value of the financial instruments booked in the Group accounts (models, parameters)

Market Supervision and validation: a complex control frameworkGeneral Overview: Functional organization & delegation principles

Page 4: Ch&cie model pricing validation 20140922_risk & finance

4

Ensure correct representation of operations in the official systems

Determine the market parameters to be used and ensure their daily contribution

Contribute to the observability assessment work

Propose modifications to the models and valuation methodologies

Supervise model implementation work

Contribute to the economic P&L validation

Are responsible for the implementation of the FO systems that are secure and that fulfil the control objectives.

Front Office

Finance

Operations

Risk

Define the adequate economic valuation methodologies and establish a reserve policy covering model, parameter and liquidity risks

Approve and review the models used by the Front Office

Draw up and maintain the “models/products” mapping

Contribute to the controls over deal representation in the systems, when no booking rules have been set

Have authority over the observability status of market parameters and products

Are directly responsible for the control of the non-standard market parameters, and are responsible for assisting Operations in the implementation of the standard parameter controls

Determine reserves.

Ensure that the deal representation in the official systems are compliant with a set of pre-defined rules

Ensure that transaction details booked by the FO that impact the economic revaluation are properly reconciled with thecontractual terms

Validate the “standard” market parameters

Contribute to the reserves calculation process (under the responsibility of RCM)

Produce, analyse and validate (substantiate) the official P&L

Contribute to the reconciliation between the accounting P&L and the economic P&L

Contribute to calculation of the Day One P&L adjustments

Ensure the accurate processing of operations (i.e. clearing and settlement, payment and cash management, confirmations)

Perform operational controls (i.e. resolution of unsettled deals, reconciliation of cash and securities movements withclearer/custodian/ broker)

Middle Office & Product Control

Back Office

Ensure the supervision of the entire Valuation and P&L Control framework (first and second level controls) through theconsolidation and analysis of the reports received from all the contributors to the Valuation and P&L Control Chain

Prepare and coordinate the monthly and quarterly meetings.

Coordinate the governance structure, namely monthly P&L and quarterly executive

Headquarter

Local Perform the first level controls that are within the Finance area, notably accounting controls;

Perform the reconciliation between accounting and economic P&Ls,

Assume the entity-specific part of the “CIB Financial Control” supervision mandate

1

2

3

4

Market Supervision and validation: a complex control frameworkFocus on mission statements (Key responsibilities)

Page 5: Ch&cie model pricing validation 20140922_risk & finance

5

Front Office

BackOffice

Finance

Risk

MISSION STATEMENTPROCESS

TRANSACTION APROVAL

DEAL EXECUTION AND BOOKING

MODELS(Initial development, implementation in

the systems and Model control framework)

RESERVES AND VALUATION ADJUSTMENTS POLICY

MARKET PARAMETERS VALIDATION

P&L PRODUCTION

Compulsory member with escalation right in dedicated committees (NPC, TAC)

Responsible for 1st level controls oncomplex deals booking

Responsible for the model control framework (approval, review and mapping)

Responsible for uncertainty or liquidityreserves valuation

Responsible for controls defined in the flowcharts of official market parameters Responsible for controls on “non standard” parameters

Validate the observability status ofparameters (for the Day One P&Ladjustments)

Model conception & implementationformally approve any new valuation model or

modification of valuation methodology following aspecific procedure (supervise back-testing andnumerical tests performed by Research/IT teams)assess the validity of the model’s theoretical

representation and the adequacy of the model tothe product to which it appliesreview the results of tests on reliability and quality

of the IT code, and has authority to ask that furthertesting is carried out andfinally authorises the use of this model for officialvaluation (go-live)

2. Model operational useis responsible for the setting and the maintenance

of the list of official (authorised) models, thatincludes the numericalconfigurations, the calibration procedure and/orset, and the official usage rules (scope of productsto which a model appliesthrough the product/model mapping)is in charge of verifying that the valuation model

used for off-systems deals is adequate (inaccordance to the product/modelmapping)Perform specific controls on deals with not yet

models or specific characteristics (reserves, limits…)

Zoom # 1

Zoom # 2

Market Supervision and validation: a complex control frameworkZoom on the risk function

Page 6: Ch&cie model pricing validation 20140922_risk & finance

6

Pricing & Validation key issues

Approach

Appendixes

CH&Cie presentation: risk management offer

Agenda

1

2

3

4

Page 7: Ch&cie model pricing validation 20140922_risk & finance

7

Market Model review

Review of MtModel

consistency & robustness

Review of Model and

pricing system

Mapping &

output analysis

Analytical review

of model results

Gap analysis of

key parameters

Dif ferences

explanation

Data

quality

Inputs /

componentsModel design

Design

benchmark

Calculation

process

Closed FormulaMonte Carlo

simulationsTrees / other …

Scenarios

review

Simulations

convergence

Market Risk

parametersOther Risk

Market direct

access

No access =>

MtModel

Partial access /

Smoothng /

interpolation

LquidityMaret

volatlity /

stress

CVA/DVACross

gamma

effect

Step 2:

Review global

methodology

Step 1:

Preliminary

diagnosis

Step 3: detailed review of a core

component

Arbitrage

Correlation

Step 2:Review global Methodology

Step 1:Preliminary Diagnostic

Step 3:Detailed review of the core components

This approach is also designed to address regulatory expectations

Pricing & validation: a vertical integration in businessCH&Cie review approach

Page 8: Ch&cie model pricing validation 20140922_risk & finance

8

Reviewing a model should encompass:

Qualitative process:

Qualitative review and management oversight

Model operating environment

Systems implementation

Data quality checks

Examination of assumptions

Quantitative process:

Review of input and parameters

Model replication

Benchmarking and hypothetical portfolio testing

Back testing and stress testing• Profit and loss attribution

The model operating environment includes:

Model documentation and its review

Review of theoretical soundness

Review of model implementation (including systems anddata quality)

Review of model inputs

Review of model assumptions, limitations and usage

Implementation and review of model controls

Environment analysis:

Vacuum of the snapshot

Heterogenity & asynchronicity

To validate a model is not strictly limited to a

quantitative review. The environment and

the internal organisation’s « fit » is

also tested

Model review functions in motionFrom a quantitative tool to a more business oriented instrument with strategic guidance

Page 9: Ch&cie model pricing validation 20140922_risk & finance

9

Is the model answering all the

bank expectations?

What is the trading strategy? What are the criteria for validating a model?

Risk of mispricing? (new model, strong assumptions, strong hypothesis …)

Very sensitive model? (Greeks and parameter sensitivities are high …)

Risk of P&L swings? Easy to Hedge or not? Very expensive to hedge?

Complex to follow or not? (change in portfolio composition / change in the underlying maturities …)

Risk of arbitrage?

No benchmark? Mark to Model? (no market price, partial quotes …)

Illiquid market? (higher bid-ask spreads…)

Instability of the model under stress conditions?

Regulatory risk? (Arbitrage in ISDA or CSA contracts …)

Capital requirement is too high? (Basel III, cash collateral requirements …)

Avoid gamma holes When volatility is high, gamma is high,

hedging is expensive Large gamma may show imperfect hedge and

possible jumps in PnL (barrier options) When gamma changes sign (spread options),

delta hedge is not possible

Monetize variance risk premia Sell implied, buy realized volatility by creating

a flat dollar gamma portfolio, go long gamma

Volatility term structure arbitrage After the crisis we expect short volatility to

decrease and long volatility to increase Sell short volatility, buy long volatility by delta

hedged straddles

Smile arbitrage Volatilities are extremely volatile, but volatility

smile is always flat Sell straddle, buy butterfly

Monetize liquidity risk premium Borrow on short-term, lend on long-term

Model review functions in motion Critical choices and model functions needs to be tested

Page 10: Ch&cie model pricing validation 20140922_risk & finance

10

Pricing & Validation key issues

Approach

Appendixes

CH&Cie presentation: risk management offer

Agenda

1

2

3

4

Page 11: Ch&cie model pricing validation 20140922_risk & finance

11

A Range Accrual is a structured product which pays a coupon based on the performance of the underlying (Equity Index, FX or interest rates…)

The Performance coupon pays an amount based on the number of days the underlying performance is within a defined Range

Sometimes, a short option (generally a Put Down & In) is added in the package; in this case, the capital is not guaranteed, but the Accrual In Rate is greater

Sometimes, the product is callable: if at any fixing date, the underlying performance is above the Range Cap, then the product is called (totally redeemed @ Capital + last coupon)

Description

Days within Accrual Range

Days out of

Accrual Range

Days out of

Accrual Range

Illustration with a Range Accrual on Libor 3M

The coupon can be modeled as the sum of • The Accrual In Rate for all the period• A series of daily short binary strangles (with Strike 1 = Floor, Strike 2 = Cap, Notional = Accrual In Rate / Total

Number of days)

Modeling

At each fixing date, the performance rate is:

Accrual In Rate & Accrual Out Rate are defined at inception, generally Accrual Out Rate equals 0. Accrual In Rate depends of Range Floor & Cap levels

PayoffNumber of days within Accrual Range

Total Number of daysNumber of days out of Accrual Range

Total Number of daysX Accrual Out RateXAccrual In Rate +

Range Accrual Description

Page 12: Ch&cie model pricing validation 20140922_risk & finance

12

There are 5 periods of one year At each observation dates (everyday), the performance used is the worst performance between S&P500, NKY

& EUROSTOXX50 performances The Accrual Range is 80%-100%, The Accrual In Rate is 7%, The Accrual Out rate is 0 The Put Down & In can be activated only at maturity (Final PDI), its activation barrier is at 80%, and its strike at

100%

Description

The coupon of one isolated period can be modeled as the sum of • The Accrual In Rate for all the period• A series of daily short binary strangles (with Strike 1 = Floor, Strike 2 = Cap, Notional = Accrual In Rate /

Total Number of days) The Call-ability adds a conditional factor on each future coupon, on the PDI, and on the Capital Funding The PDI adds a jump of (strike - barrier) at its barrier

Modeling

At each fixing date (each year): • Performance rate (chart below): The performance was within the

Range n days on a period of N days, the coupon is: Accrual In Rate * n / N

• Call-ability: If the performance is above the range, then the product is called (totally redeemed @ Capital + Coupon of current period)

Payoff

At maturity, if the product has not been called,

Last Coupon: Performance coupon as for other periods

Final PDI 80%/100% (chart below): If the Performance is under 80%, the Put Down & In is activated, then the product redeems the capital multiplied by the underlying performance (if the performance is -40%, the note redeems 60% of the initial capital)

An example: a 5 years 7% Callable Range Accrual 80%-100% with 1 Final PDI 80%/100% on the worst performer of (S&P500, NKY, EUROSTOXX50)

Page 13: Ch&cie model pricing validation 20140922_risk & finance

13

Pricing & Validation key issues

Approach

Appendixes

CH&Cie: Focus on the Risk Management offer

Agenda

1

2

3

4

Page 14: Ch&cie model pricing validation 20140922_risk & finance

14

CH&Cie Risk Management offer (1/4)From managing risk processes, to measuring risks and establishing strategic guidance

Strategic

guidance

Measurement &

validation

Processes &

organisation

Risk

Management

1

23

• Helping to making high-level decision (CVAdesk implementation etc…)

• Defining risk appetite in accordance withthe business strategy & development

Strategic guidance

Measurement & Validation• Quantifying risks and measuring

impacts on a business level

• Validating models and developing advanced quantitative techniques

Processes & organization• Reviewing risk management

processes

• Establishing monitoring procedures

• Organizing and defining risk governance and follow-up

Page 15: Ch&cie model pricing validation 20140922_risk & finance

15

1. Finance 2. Pricing3. ALM / Liquidity

4. Credit Risk5. Market Risk

6. Operat. Risk

7. Business & Strategy

8. Customer relationship management

1.1 ICAAP / Pillar 2

1.2 Economic capital

1.3 Capital budgeting / RAPM

1.4 P&L and budget forecasting

2.1 Standard & Complex Models

2.2Instrument pricing

2.3 Pricing Parameters control

3.1 Basel III : LCR, NSFR, liquidity

3.2Securitization SPV, collat. manag.

3.3 Gap : CF patterns, survival horiz

3.4 Dynamic modeling

4.1 Basel II: PD, LGD, EAD, CCF, UL, RWA

4.2 Basel III, CVA, CCP, Capital

4.3 Solvency II : capital

4.4 Provision specific, collective

4.5 Stress & back testing

5.1 Classic & stress VaR, CVar

5.2 Risk reserves

5.3Sensitivities Modeling & Calculation

5.4Incrementaland liquidityrisk

6.1 Fraud detection

6.2 AMA models

6.3 Rogue trading

7.1 Strategy guidance and decision

7.2 Brand notoriety, reputation

7.3 Process optimization

8.1 Credit granting models

8.2 Portfolio scoring

8.3Marketing and targeting

8.4 Data mining and desctriptive statistics

CH&Cie Risk Management offer (2/4)A large scope of intervention with expertise, experience and benchmarking at the heart of our strategy

0. Advanced Modeling, experience, expertise, benchmarking

Please, specify the subjects you are interested in, by checking the orange boxes

Legend

Business intent

Regulatory intent

6.4Operations structuringcontrol

Page 16: Ch&cie model pricing validation 20140922_risk & finance

16

CH&Cie Risk Management offer(3/4)Modeling as an integrated business tool: a cross-disciplinary skills and decision-making facilitator tool

Modeling as a transversal tool

Risks1

• Market : VaR computing, volatility,liquidity, valuation

• Credit : Basel II parameters,Provisioning, stress, back testing

• Operational : fraud, rogue trading...

Finance2

• Manage Assets and Liabilities

• Manage Economic capital (ICAAP)

• Simulate P&L impacts

• Capital Budgeting : RAROC etc…

Business3

• Optimize operating model

• Adapt marketing (CRM)

• Scoring and targetingcustomers

Strategy4

• Build business strategy

• Monitor reputation

• Arbitrage between risk takingand business developement

Modeling allows to anticipate, prevent, detect, measure, test, develop and decide… It is a powerful tool that requires a specific set of skills and knowledge

Page 17: Ch&cie model pricing validation 20140922_risk & finance

17

CH&Cie Risk Management offer(4/4)Modeling techniques and requirements: the work tools

Data analysis1

• To give a quantitativeperspective of aspecific context or forproblem detections(by analysing data)

Main objectives

Simulation2 Solving3 Prediction4 Methods5

• To validate hypotesisand / or find the bestoption of a specificstrategy

• To give a closedformula of a specificproblem

• To give an estimate ora prediction (estimedprobability of an eventto happen undercertain hypothesis)

• To define and design aquantitativemethodoloy forstrategy purposes orbusiness decision

• Data Mining• Statistics

Underlyingtechniques

• Monte Carlo simulation

• Bayesian networks• Fuzzy logic / Expertise

• Mathematics• Statistics

• Probability• Statistics

• Benchmark• Experience/ Best

practices

• Fraud detection• Portfolio analysis• Correlation analysis• Dashboard / reporting• Marketing …

Illustrations • Capital planning• Strategic plan

forecasting• Pricing• Stress testing …

• RWA Calculation• Pricing• Marketing• Valuation (firm

value)…

• Risk parameter estimation (PD, LGD, EAD)

• VaR / Credit VaR …

• CVA desk implement.• « Cost of risk »

hedging policy• Choice among

different approaches…

AAA

AA

A+

A-

BBB

BB+

BB-

B

CCC

DX-

200

400

600

800

1 000

1 200

20

11

20

12

20

13

20

14

20

15

2 0

20

2 0

30

2 0

40

2 0

50

2 0

60

2 0

70

2 0

80

2 0

90

2 1

00

Rating

Number of clients

Maturity

Profile analysis

i

ii

i

yp

yYZP

1

)(

1

1

Markov

Models

Regression models

Vintage analysis

Binomial Tree

Actuarial models (Beta calibration)

Statistical

Models

Loss Calc

Others...

External

Models

Recovery

Assessment models

Page 18: Ch&cie model pricing validation 20140922_risk & finance

MONTREAL

202 – 1819 Bd Rene

Levesque O.

Montreal, Quebec,

H3H2P5

PARIS

20, rue de la Michodière

75002 Paris, France

NIORT

19 avenue Bujault

79000 Niort, France

NEW YORK

1441, Broadway

Suite 3015, New York

NY 10018, USA

SINGAPORE

Level 25, North Tower,

One Raffles Quay,

Singapore 048583

HONG KONG

905, 9/F,

Kinwick Centre 32

Hollywood Road,

Central, Hong Kong

LONDON

50 Great Portland Street

London W1W 7ND

UK

GENEVA

Rue de Lausanne 80

CH 1202 Genève,

Suisse