Transcript Aptivaa Approach for Market Risk IMA (Internal Models

Analytics
Consulting
Solutions
Transition to IMA: How to
ensure a smooth ride?
211h July 2010
Business Case
2
Business Case -IMA Approach
IMA is not merely a regulatory compliance framework; by taking a broad business
perspective on the Accord, a bank can capture a series of strategic opportunities
Strategic Opportunities in IMA Adoption
Potential Threats of IMA Adoption
A. Capital Increase: Banks expect a significant
increase in capital requirements under IMA
A. Reputation : Enhanced perception of the Bank
as a low risk and regulatory compliant bank in
order which improves/maintains the bank’s cost
of funds and share price
B. Relatively Small HFT portfolio: For Banks with
relatively small HFT portfolio compared to AFS, it
may not make business sense for migrate to IMA
from ‘cost benefit analysis ‘
B. Better Risk Management: A more precise risk
weighting of assets
C. Capital Management: Internal efficiencies and
disciplines in Risk reporting and monitoring
would lead to changes in product mix and
pricing.
D. Rating: Banks may decide to grow via
acquisitions or divest a portion of their assets, in
order to maximize the benefits of a more efficient
capital management model
E. Competition - Driven by the Peer group
3
IMA Framework
4
Qualitative Requirements for IMA approval
Independent
Risk Control
Unit
 The bank must have a Risk Control Unit that is fully independent of business units that generate
market risk exposures.
Documentation
 The bank must have well documented policies for Trading Book, Market Risk, ALM Procedures
 Also, the bank must maintain an MRM Dossier to record model specification, changes and
updates.
Integrity &
Accuracy
 The bank must be able to demonstrate that it has a conceptually sound risk management system
that is implemented with integrity
 The bank must have a proven track record of measuring risks with reasonable accuracy.
Skilled-Staff
Use-Test
 The bank must have a sufficient number of skilled staff using sophisticated methods in trading
area, market risk control, validation and internal audit.
 The risk estimates produced must be closely integrated with the risk management process.
 The risk measurement system should be used in conjunction with internal trading and exposure
limits.
5
Quantitative Requirements for IMA approval
VaR parameters
Data
Stress and Back
Testing
Empirical
Correlations
Non-Linear Risks
 VaR must be computed on a daily basis using a 10-day holding period and confidence level of
99%.
 No particular model prescribed, but the model in use should be able to capture all the
material risks
 Historical Time horizon for calculating VaR will be constrained to a minimum length of one
year.
 Banks must update their data sets no less frequently than once every three months
 A rigorous Stress Testing exercise should be in place to supplement the risk analysis.
 The bank must conduct a rigorous Back Testing to carry out an ex-post comparison of the risk
measure generated by the model against actual daily changes in the P/L as well hypothetical
changes.
 VaR models should ideally capture correlations across broad risk categories spanning Market
Risk (Interest Rate, FX, equity Price, Commodity)
 Models must accurately capture the unique risks associated with options with suitable models
 For banks using advanced derivative products, the models must capture the non-linear risks
like gamma, vega.
6
Typical Implementation Approach
Phase 0
Scoping Study
1 -2 wks
Key Methodologies / Tools
Establish
Project Scope
Understand:
Bank structure
Programme
structure
Key decisions
made
Phase 1
Gap Analysis &
Planning
Phase 2
Framework Design
& Specification
4 - 6 wks
Phase 3
Implementation Phase
6 - 8 wks
Policy &
Procedures
Basel II Gap
Analysis :
Existing vs.
Target State
Data Validation
Project Planning/
Resource
Management
Construction/
Selection of Market
Risk System
Define projects
to fill the
identified gaps
Configuration of
Pricing/VaR
Models
24 -50 wks
VaR and Stress Testing Models
Back-Testing*
Model Refinement
Workshops & Training
Detailed Documentation
Application to RBI
Business Transformation ( Training & Communication )
* The
implementation of the back testing program should begin at least six months before the bank decides to make an application to
RBI for approval.
7
Market Risk Governance Under IMA
 IMA guidelines make a strong emphasis on the Organizational Structure and Quality of Governance for
Market Risk
 As per the guidelines, BoD and senior management should be proactively involved in Market Risk
Control.
A typical organizational design of the market risk function of a bank complying with Basel II framework for
Market Risk would have the following ‘Independent’ entities.
Market risk organizational structure
Front
Office/Trading
Unit
Middle
Office/Risk
Control Unit
Model
Construction
Unit
This desk
comprises the
trading desk
and their
immediate
supervisors
The unit is
responsible for
design and
implementatio
n of the bank’s
risk models,
Stress and
Back-testing.
If the bank’s
model are built
in house, this
unit should
comprise of
staff who are
not involved in
model
validation or
internal audit.
Model
Validation Unit
This unit
comprises of
staff who take
care of
validation and
were not
involved in
construction at
all.
8
Back office
This unit
ensures the
correct
recording of
transactions
and funds
transfers.
Internal Audit
This unit is
responsible for
carrying out an
independent
review of
activities of
trading unit
and Risk
Control Unit.
Design and Methodology
9
Current VaR Practices
Currently, banks in India calculate capital charge for Market Risk using SMM and use different VaR
models( mostly Historical Approach) for calculating Market Risk for Internal Purposes.
TYPES OF VaR Models
Historical
 Non-parametric method that involves re-valuing portfolios against a set of
historical scenarios assuming they are a good representation of all the
possibilities between today and tomorrow.
Monte-Carlo
Simulation
 Parametric Method similar to historical simulation, but instead of using
historical scenarios, it generates them randomly assuming that portfolio
returns follow a Normal Distribution
VarianceCovariance
 Parametric Method involves calculating VaR analytically by
making assumptions about return distributions and by using
variance- covariance
A typical Market Risk Distribution for a Sample Portfolio
10
Selection of VaR Methodologies
Features
Historical Approach
Monte Carlo
Simulation
Variance-Covariance
Shape
Actual
Usually Normal
Normal
Use Correlations
No
Yes
Yes
VAR precision
Poor with short window
Good with many iterations
Poor
Ease of computation
Easy
Intensive
Intermediate
Accuracy
Yes
Yes
Poor for non-linear
dependencies
Communicability
Easy
Difficult
Easy
VaR analysis
Easy
More difficult
Easy
Yes
Yes
No
Historical events maybe
irrelevant in the present
context
Model risk
Needs regular Updation
of variance-covariance
Matrix
Distribution
Implementation
Capture Non-Linear Risks
Other s
11
Stress Testing
TYPES OF STRESS TESTS
Scenario Building
-Historical
Analysis
Sensitivity
Analysis
MechanicalSearch
‘Reverse’ Stress
Testing
 This type uses scenarios from recent history, such as the 1987 equity crash, 2007 subprime crisis and simulate the effect on P/L of repeats of past historical events.
 This type uses pre-defined or regulator prescribed scenarios that have proven to be
useful in practice. Example, change in equity price by x%, yield curve shift of y-basis
points.
 This type uses automated routines to generate over prospective changes in risk factors,
and report the worst effect on P/L.
 Reverse Stress-tests require a bank to assess scenarios and circumstances that would
render its business model unviable, thereby identifying potential business vulnerabilities. It
starts from an outcome of business failure and identifies circumstances where this might
occur.
12
Pitfalls in Currently employed Stress Testing Methods
Subjective
Nature
Difficult
Interpretation
Lack of
Sophistication
Stress Testing
tests are
unavoidably
subjective
because they
depend on
scenarios
chosen by the
stress tester.
Hence, its value
depends on the
choice of
scenarios and
on the skills of
the modeler.
The results of
stress tests are
difficult to
interpret
because they
give us no idea
about the
probabilities of
the events.
Methodology of
stress testing is
still in its infancy,
and the
approaches
commonly used
are open to
objection. No. of
risk factors that
can be stressed
simultaneously
is capped.
13
Integration
With
VaR models
VaR estimates
which are
probabilistic in
nature cannot be
integrated with
loss estimates
generated by
Stress Tests
under different
scenarios.
Back Testing
The Stress-Test
procedures are
very difficult to
back-test as
hypothetical
stress scenarios
cannot be
“validated”
based on actual
events.
Model Validation under IMA
Validation must be carried out independently by internal and external auditors before
RBI Validation
Model validation consists of assessing five major
components:
 Governance & Oversight
 Data integrity
 Assumptions
 Methodology
 Back testing
14
Documentation
17
Documentation
For obtaining RBI approval for the IM risk models of banks, the following documents are to be
submitted:




Request for RBI approval of the IMA model
Internal audit report of the model
A Market Risk (MR) file
MR Model Dossier
MR File Content
MR file describes the internal
model, the risk management
control system and substantiates
the compliance with the
quantitative and qualitative
requirements of these guidelines.













Scope of Application of the Model
Description of Exposures
Estimation of Regulatory Capital
Policies and Organization
Market Risk Measurement System
Stress Testing Programme
Back Testing Programme
Technological Environment and Information Integrity Controls
Limits Structure/Information Systems/ Databases Employed
Operational Manual and Input Tables of the Market
Risk Calculation
Internal Audit Report
Other Independent Assessments
Future Developments
18
The MRM Dossier
The content of this document is similar to the content of the MR file. However, the difference between
the two lies in the fact that the MR file is submitted at the time of application to the RBI, whereas the
MRM Dossier is maintained and updated on a regular basis.
Apart form the various components of the MR file, the following information must also be contained in
the MRM Dossier:
MRM Dossier


The purpose of the MRM Dossier
is to keep a record of the details
of the model and the
changes/refinements, if any,
made in from time to time.





Full technical specifications of the model
RBI approval for the model and subsequent changes,
if any, made to the model along with the conditions subject to
which the approval has been granted.
Complete details and record of subsequent changes,
if any (such as new products covered, modifications to the
score of the model, revision of sources of external data,
modifications in the applications, organizational changes etc.)
in the operation of the approved model.
Authors responsible for the contents, date updated
Detailed Stress Testing and Back Testing Results
Uses to which the VaR is put within the bank
Weaknesses identified in the model
19
Snapshot of Dossier

MR File Volume: typically 200-300pages (A4 Word)

MRM Dossier Volume: typically 150-200 pages (A4 Word)
20
Documentation for Market Risk policy
Typically 50-100 pages (A4 Word)
21
System/IT
22
Risk Technology Architecture
Risk Technology architecture would be driven by the functional Risk architecture design and implementation at
Group and Entity level. A generic risk technology architecture (illustrative) highlighting the key technology and
systems components is presented below.
Source
Systems
1
ETL
Tool
2
Oracle
/SQL /
DB2
Retail
Operationa
l Systems
Wholesale
Operationa
l Systems
Finance
Systems
Data
Integrati
on
(ETL)
Risk
Databas
e / Data
Warehou
se
3
4
Credit Risk
Operational Risk
Collateral & Limit
Mgt
Operational risk
mgt solution
Credit Workflow
Operational risk
modeling
Rating Engine
Compliance
Tracker
Risk
Reporting
&
Analytics
Regulatory
Reporting
ERM Solution
Components
8
Market Risk
Capital
Management
Market Risk
Analytics
Regulatory Capital
Engine
ALM & Liquidity
Risk
Economic Capital
Modeling
Counterparty
Credit
Budgeting and
Forecast
5
7
Risk
Reporting
9
Risk
Analytics
6
23
Timing of System Initiatives
Risk IT Architecture
NO CHANGE
Rating
systems
UPGRADE
Risk Modelling
System
UPGRADE
NO CHANGE
Economic Capital
•
Finance (Capital
Budgeting)
Retail Risk
Systems
NEW (W)
201X Q1
Multiple systems are used to
source the risk data and then
run through series of data
transformations and
integration layers for Pillar I
/ECAP calculations.
Global
Counterparty
Limit
Management
(Fermat)
•
NEW (W)
201X Q1
UPGRADE
NO CHANGE
Basel II
Calculator
(Bancware)
Credit
Workflow
(Fermat)
Operational
Risk
Management
(OpenPages)
ETL (Data
Transformations
Management)
NEW (W)
201X Q1
Enterprise
wide Risk
Dashboard /
Reporting
components* to provide for
an enterprise wide risk
dashboard / reporting
mechanism
Pillar II Risk Data
Warehouse
Stress
Testing
•
NEW (W)
201X Q1
Market Risk
(Algorithmics)
Pillar II
Reporting
* The diagram outlined is indicative based on our understanding and may/may not
reflect the current or proposed architecture at the Bank.
Build
Buy
N/A
Additionally, the Risk
Technology Architecture
must be interfaced with
Finance Systems and the
impending ERM
With some systems already
bought/built/implemented
and or under implementation
(either bought or being built),
some interim adjustments /
upgrades may be made in
the architecture to ensure a
practical transition to the
target risk architecture.
24
Challenges for the Supervisor

No clarity on data period for calculating ‘Stressed VaR’ makes it difficult to
assess its value objectively .

Under Stress-Testing, scenario building depends on the choices made by the
modeler and his skill-set which could make it difficult to compare its results with
that of Banks of similar size.

For banks with spreadsheet based IMA models, it is difficult to maintain audit
trails and controls

Back-Testing results could be meaningless if the portfolio of a bank changes
drastically and frequently. In that case Back Testing has to be done more
frequently depending upon the nature of the portfolio.

Staff-skill set needs to be diverse in order to review and validate different and
eclectic models used by the Banks as no particular method is prescribed.
25
Thank You
NORTH AMERICA
EUROPE, AFRICA AND THE MIDDLE EAST
INDIA
ASIA AND AUSTRALASIA
Aptivaa Inc.
100 Park Avenue
Suite 1600,
New York, NY 10017
USA
Aptivaa Consulting UK Ltd.
Three Tuns House
109 Borough High Street
London. SE1 1NL
UK
Aptiva Consulting Solutions P. Ltd.
Trade Star, ‘B' Wing, 5th Floor
Andheri-Kurla Rd,
Andheri (East), Mumbai - 59
INDIA
Aptivaa Consulting (S) Pte Ltd.
Level 31
Six Battery Road
Singapore 049909
Tel: +1 212 984 0782
Tel: +44 20 7096 9888
Fax: +44 20 7744 6259
Tel: +91 22 4083 8600
Fax: +91 22 4083 8601
Tel: +65 6725 6241
Fax: +65 6725 6242
Confidentiality clause
This document is confidential. No part of it may be circulated or reproduced outside without express approval of Aptivaa Consulting.© Aptivaa Consulting 2010.