Valuation, Model and Data Risk Management - GARP

“Only the curious will learn”

Valuation, Model and Data Risk Management GARP Caribbean Chapter Meeting, March 28th 2008

By Philippe Carrel, Reuters, Executive Vice President [email protected]

"It was the failure to properly price these risky assets that set off the tidal wave of risk contamination" . Alan Greenspan, speaking at Reuters on October 1st 2007

2

Growth, competition lead to accelerated innovation The quest for alpha leads to innovation • Hedge Funds / Prop trading develop absolute return strategies • Cross-asset strategies • Combination of loans, securities, OTC instruments,hybrids Assumptions, Correlations, Proxies, Models Risk increasingly managed from front office

Fast growth, competitive markets, accelerated innovation • Securitisation boom (US$1.2 tn in 2006) • Liability driven structuring Assumptions become usual practice Liquidity produces liquidity

3

Of the dangers of uncontrolled automation

Uncontrolled automation lead to systematic risk • Program trading • Automated arbitrage and order generation • Implied volatilities and correlations • Automated feed of reference data Risk is increasingly managed from front office, model driven, assumption based

Loss of confidence lead to systemic risk • Sub-Prime • Credit crunch Catalysts invalidate assumptions in the short run which leads to panic

4

Greed, Fear and Regulations Regulators’ obsessive focus on risk… • Basle II Triggered massive transfers of assets to special purpose vehicles Freed capital

•Mifid/NMS Multiplication of trading venues Data repositories, analytics platforms

• UCITS III Generalise use of derivatives and OTC instruments Valuation-based compliance

• SFAS 157 Defines fair value as sell value Establish 3-level hierarchy based on valuation

…or on valuations actually. 5

Innovation Competition

A new risk class

Market Risk Credit Risk

Operational Risks

Data & Valuation Procedures

6

Data + Models + Processes + Control = Valuation risk For…

Data Acquisition

Data Processing

..we need to… Orders Static Data OTC Data Execution Position/PL Analytics

Identify Assess Mitigate

Data Distribution

7

..the following risks :

Pricing Valuations Contribution

Reporting

Portfolio Risk Performance

Compliance

Regulatory Audit control Reporting

Monitor Report

Data Risk Valuation (Process) Risk Model Risk Sensitivities Collateral Liquidity

(Mkt/Crdt)

Valuation risk Data • Prices • Reference data • Models and hierarchy • Formats / Taxonomy

Processes

Which data for which purpose? Workflow analysis

How does the firm use data, rely on processes and assumptions?

• Compatibility of models and data, consistency in use • Data quality management (rating on availability, reliability, updates) • Data hierarchy (dependencies on imported, existing, calculated data) • Process hierarchy (critical path, dependence on assumptions) • Workflow analysis (potential domino effect) • Vulnerabilities (latency, breakdown) • Connectivity (Synchronicity, external communications)

Organisation

Has the firm established valuation risk as part of its risk culture?

• Sponsorship, hierarchy, culture • Governance • Dashboard / warnings, alerts • Alternative procedures 8

Golden copies and set-in-stone analytics…

Exchanges

OTC Markets Loans

OTC Derivatives Derivatives

OMS Providers

FX

FX OTC deriv Futures

Orders

Securities

Execution Venues

Fix Inc

CDS

Futures & Opt

Brokers

Fix Inc Deriv, Equ, CFDs, Cmdts

Position Analytics and Valuations Real-time export Front Office

Back Office

Admin & Accntg

Asset Mgt

Adhoc Reporting for Communities

9

Custody

..quickly reveal their own limitations. Cash / Fx / Fwd / Opt / MM / STIR Deriv / Bills / Credit / Fix Inc/ MBS / CDOs Futures / Fut Opt / Equities / CFDs / OTC Deriv Metals / Energy / Softs

Exchanges

OTC Markets

Electronic Execution Market Makers Distributors OMS Providers Brokers Others

Trade Entry

Trade blotter

Order Processing Trade allocation

Exception Mgt

Input verification

Inventory Analysis

Book Structure

Position Management

Position Analysis

Funding

Amendment forms

Drill-Down Scenario Builder

Order Generation

Parameters Collateralizing deals

Collateral Management

Instrument Requests

Ticket/Pricing/Order New bond forms CA Event forms

Collateral reports

OTC inst forms Instrument search Instrument analytics

MARKET DATA

Position Record

Notifications

Derivatives Pricing

Data Management

Curves / Surfaces

Hedge effectiveness

End of Day/Month Maturity Option expiry

Network Communications

Export to Admin / PB

Exception Mgt

Terminations Cash events

Trading Reports

Operational Reports

End-of-day Reports

Compliance Reports

Reporting 10

Trading

Operations

3rd party reports

Output RM

Communications

Processes

Open models, dynamic, adaptable workflow… Input Services

Data Acquisition

Input & Visualisation Calibration Pricing Pre-Trade Analytics

Data Processing

Position Analytics Portfolio Reports P/L & Risk

Data Distribution

Scenario-based analytics Limits & Compliance

Issuer Data T&Cs Corporate actions Market Data

Valuations

Interactive Platform

Projections Yield Curves Risk aggregation

Vol Surfaces

Performance

Correlations

Decomposition

Ratings

VaR & Stress Test

Interactive Services

Scenarios Portfolio modelling X-Asset Aggregation

Margin & Collateral Back-Office Records History & Audit

11

Communication Services External connections

Risk & Compliance Reports

..dynamic adaptable processes,… Workflow and hierarchy • Contingencies in processes • Data dependencies • Defining Key Risk Factors (feed, curves, surfaces, matrices, diffusion model) for concentrations

Scenario based risk analysis • Sensitivity to chosen factors, observable input, economic factors • Interdependence of risk factors (volatility, spreads, correlations) • Impact on collateral, covenants (rule-based) • Multiple model approach (highly recommended by most professional associations) • Volume processing, scalability • Connectivity, external contingencies

Triggers and alerts • Establish triggers (intra-day vol, price gaps, default event) • Measure impact immediately • Alternative procedures

Alternative procedures • Swappable, adjustable analytics • Alternative data sets • Alternative processes • Position adjustments 12

..interactive data models… will make the firm more agile. Insights • Huge volumes to be analysed and visualised almost instantly • Multiple views, interactive queries, filtering • Multiple data sources –aggregated an reconciled automatically • Multiple pricing models • On-demand variables and risk factors aggregation • Permanent compliance check against investment policies, risk limits

Open Model Approach • Multiple (alternative) models • Scenario-based, cross-asset

Combining the management of data, models and processes into a set of adaptable and interchangeable procedures for pricing, valuing and reporting, P/L, exposure and performance. How agile a firm can be in adapting to a new business environment.

13

Organisation and corporate culture Structure • Risk culture acknowledging Valuation Risk as a standalone risk to be managed • Risk committee • Hierarchy, Empowerment • Independence

Policy/Strategy • Governance and acountability (process owners) • Taxonomy • Hierarchy, Empowerment

Audit and control • Openess • Disclosure of data, results, methodologies • “Democratised access to data (user interface)

14

“As far as mathematics refer to reality they are not certain, as far as they are certain they do not refer to reality” Albert Einstein

Valuation, Model and Data Risk Management GARP Caribbean Chapter Meeting, March 28th 2008 By Philippe Carrel, Reuters, Executive Vice President [email protected]