3rd Annual RISK AMERICAS, May 12-13, Marriott Downtown, New York City

Presenter: Anshuman Prasad, Director, Risk and Analytics

NEW DIRECTION FOR MARKET RISK:

ISSUES AND CHALLENGES

May 12-13, 2014

Agenda

Market Risk Management: Outstanding Issues

Impact of Regulations

– Measures of Market Risk

– Regulatory Stress Tests

Modeling Challenges

– Valuation of Illiquid Instruments

– Illiquidity Discount

– Computational Trade-offs

Data Management Issues

Trends in Market Risk Systems

Organizational Culture

2

Market Risk Management: Outstanding Issues

3

Reliability of various measures of risk

Validation and back-testing of models

to reduce model risk

Trading vs. banking book boundaries

Regulatory stress tests

Regulatory Issues

Silo approach to risk management

Limited role of risk managers in

decision making

Narrow, compliance-focused

approach

Organizational Culture

Valuation of illiquid products

Correlation under stressed

environment

Model calibration

Tradeoffs: speed vs. accuracy

Incomplete and/or inaccurate data

Lack of unified data processing

model

Limited involvement of business

Real-time computation, lack of

integrated systems

System and Data Issues

Modeling Challenges

Impact of Regulations

4

Newer regulations impacting market risk management frameworks

Fundamental Review

of the Trading Book

(BCBS 219)

Regulation

Stress Testing

(Dodd-Frank, CCAR)

Model Risk

Management

(OCC SR-11/7)

Difficult to move assets between the trading and banking

books

Shift to Expected Shortfall from VaR

Varying liquidity horizons

Impact

Handling granular market shocks

Requires calculation of various parameters under stress

Widened the definition a “model”

Enhanced model validation and model governance framework

Emphasis on rigorous testing, model documentation and

quantifying model uncertainty

Data aggregation and

Reporting

(BCBS 239)

Early warning systems

Comprehensive and granular reporting

Overhaul of systems and processes

Measures of Market Risk

5

There is no single measure of risk that can provide a consistent view of risk

across market participants and regulators

Simple, established

Single view of risk

Pros

Conservative

Size and likelihood

of losses

Reduces cyclicality

Easy to interpret

Simple modeling

procedure

Lacks coherence

Extreme tails

Inconsistencies

Cons

Difficult to back-test

Prone to

optimization errors

Can be misleading

Cannot be

aggregated

Regulatory reporting

Setting risk limits

Key Use

Regulatory reporting

Economic risk

capital

Key measure used

for hedging risk

Setting risk limits

Value at

Risk (VaR)

Measure

Expected

Shortfall

(ES)

Sensitivities

Regulatory Stress Tests (CCAR, EBA)

Regulators provide a varying degree of granularity in their market shock

variables

Some of the key requirements in dealing with these stress tests from a

market risk perspective include

– Comprehensive understanding of risk drivers

– Customized scenario expansion models to expand regulatory scenarios

– Robust input-output templates to ensure quick turnarounds

– Organizational focus and coordination across functions

Real GDP Growth

Unemployment

Equity Market

Index

House Price Index Str

ess

ed

: M

ac

ro v

ari

ab

les

Econometric

Modeling

Credit Spreads

IR Term Structures

IV Term

Structures/Skews

Correlations

Valuation Models

Market Risk

Modelling

CCP Exposure

Modelling

6

By discounting cashflows at

future dates

Description

Closed-form pricing for low

factor/low dimension

models; path independent

Closed-form pricing for low

factor/low dimension

models; path dependent

IRS, Bonds, Floaters

Type of Products

First generation structures

Caps/Floors, CBs, TS

models

Simulation pricing for high

factor/high dimension

models; path dependent

Highly complex structures

Modeling Challenges: Model Calibration

7

Discounting

Technique

PDE-based

Lattice-based

MC Simulation

FASB valuation categories are based on complexity – Level 1/2/3

Level 3 assets are priced using in-house valuation models

Models built can only be as good as the underlying data !

Modeling Challenges: Illiquidity Discount

Driven by stress testing requirements, initial steps are being taken towards

better management of the market impact of illiquidity; measures include

– Checking for variability in liquidation times for illiquid positions

– Setting up processes for measuring impact of trade volumes on market prices

– Monitoring widening bid-ask spreads, specially in times of adverse market conditions

CCPs started for most liquid OTC derivatives; liquidity adjustment

challenges seen in more exotic products

Methods for incorporating the illiquidity discounts in OTC valuation models

8

Adjustment of bid-ask spread for specific market risk

parameters

Determine lower and upper limit price by relying on

reasonable limits to trade performance

Bid-ask

Spread

Adjustment

Bounding

Approach

Computational Trade-offs

9

Choice between delta-based methods and the full revaluation method

Most banks still use sensitivity-based approaches due to computational

challenges associated with the full-revaluation approach

Newer technology (GPUs, multi-core processors) helping reduce

computation times and trade-offs

Sensitivity Based

Approaches

Ease of implementation

Approximations (Taylor

expansion/grids)

Full Revaluation Approach

Robust

Computationally intense

CONTROL

AGILITY

Data Management Issues

10

Increased regulatory focus on providing granular and frequent analysis of

real time data

Traditional Extract-Transform-Load (ETL) processes not amenable to real-

time computation of risk. Alternatives may include:

– ELT: Data transformation after the data load, includes data virtualization

– Data Federation: Single virtual view without actually moving the data

Typical Data Management Processes

Trade Data

Market Data

Extract

Validate

Enrich

Transform

Reports

Ad hoc Queries

Load

Trends in Market Risk Systems

11

Increase ‘risk velocity’ and ‘risk management’ clock-speed

Automation of stress testing and reporting framework

GPU-based farms enabling move to full revaluation

approach

Measure

Cloud Computing, a natural solution for higher data

capacity needed for granular regulatory reporting

Service-based architecture with detailed data captured at

trade level

Flexible data capture and storage solutions

Timeliness and

Accuracy

Data/Reporting

Attribute

Comprehensive

ness

Adaptability

Organizational Culture

12

Front office alignment

– Financial incentives being aligned with risk-informed performance indicators

– RWA consumption by trade becoming an important statistic

Model validation and model development functions being strengthened

– Investments in internal/external resourcing

Robust middle office and IT processes becoming a differentiator

– Cost of funding (CVA, FVA etc) getting incorporated

– Granularity/Speed/Flexibility in risk calculations and reporting

Organizational structure and incentives are getting better aligned to market

risk management objectives

Market risk management will need to be more tightly integrated with strategic

objectives, diversity of business and level of complexity

www.crisil.com/gra