J.P.

Morgan/Reuters
TM

RiskMetrics Technical Document

Fourth Edition, 1996

New York
December 17, 1996

Morgan Guaranty Trust Company

Risk Management Advisory
Jacques Longerstaey
(1-212) 648-4936
[email protected]

J.P. Morgan and Reuters have teamed up to enhance RiskMetrics . Morgan will continue to be
responsible for enhancing the methods outlined in this document, while Reuters will control the
production and distribution of the RiskMetrics data sets.
Expanded sections on methodology outline enhanced analytical solutions for dealing with nonlinear options risks and introduce methods on how to account for non-normal distributions.
Enclosed diskette contains many examples used in this document. It allows readers to experiment
with our risk measurement techniques.
All publications and daily data sets are available free of charge on J.P. Morgans Web page on the
Internet at http://www.jpmorgan.com/RiskManagement/RiskMetrics/RiskMetrics.html. This
page is accessible directly or through third party services such as CompuServe, America
Online , or Prodigy.

This Technical Document provides a detailed description of RiskMetrics , a set of techniques and data
to measure market risks in portfolios of fixed income instruments, equities, foreign exchange, commodities, and their derivatives issued in over 30 countries. This edition has been expanded significantly from
the previous release issued in May 1995.
We make this methodology and the corresponding RiskMetrics data sets available for three reasons:

Reuters Ltd
International Marketing
Martin Spencer
(44-171) 542-3260
[email protected]

1.

We are interested in promoting greater transparency of market risks. Transparency is the key to
effective risk management.

2.

Our aim has been to establish a benchmark for market risk measurement. The absence of a common
point of reference for market risks makes it difficult to compare different approaches to and measures of market risks. Risks are comparable only when they are measured with the same yardstick.

3.

We intend to provide our clients with sound advice, including advice on managing their market
risks. We describe the RiskMetrics methodology as an aid to clients in understanding and evaluating that advice.

Both J.P. Morgan and Reuters are committed to further the development of RiskMetrics as a fully
transparent set of risk measurement methods. We look forward to continued feedback on how to maintain the quality that has made RiskMetrics the benchmark for measuring market risk.
RiskMetrics is based on, but differs significantly from, the risk measurement methodology developed
by J.P. Morgan for the measurement, management, and control of market risks in its trading, arbitrage,
and own investment account activities. We remind our readers that no amount of sophisticated analytics will replace experience and professional judgment in managing risks. RiskMetrics is nothing more than a high-quality tool for the professional risk manager involved in the financial markets and
is not a guarantee of specific results.

RiskMetrics Technical Document

Fourth Edition (December 1996)
Copyright 1996 Morgan Guaranty Trust Company of New York.
All rights reserved.
RiskMetrics is a registered trademark of J. P. Morgan in the United States and in other countries. It is
written with the symbol at its first occurrence in this publication, and as RiskMetrics thereafter.

Preface to the fourth edition

iii

This book
This is the reference document for RiskMetrics . It covers all aspects of RiskMetrics and supersedes all previous editions of the Technical Document. It is meant to serve as a reference to the
methodology of statistical estimation of market risk, as well as detailed documentation of the analytics that generate the data sets that are published daily on our Internet Web sites.
This document reviews
1.

The conceptual framework underlying the methodologies for estimating market risks.

2.

The statistics of financial market returns.

3.

How to model financial instrument exposures to a variety of market risk factors.

4.

The data sets of statistical measures that we estimate and distribute daily over the Internet
and shortly, the Reuters Web.

Measurement and management of market risks continues to be as much a craft as it is a science.

It has evolved rapidly over the last 15 years and has continued to evolve since we launched
RiskMetrics in October 1994. Dozens of professionals at J.P. Morgan have contributed to the
development of this market risk management technology and the latest document contains entries
or contributions from a significant number of our market risk professionals.
We have received numerous constructive comments and criticisms from professionals at Central
Banks and regulatory bodies in many countries, from our competitors at other financial institutions, from a large number specialists in academia and last, but not least, from our clients. Without
their feedback, help, and encouragement to pursue our strategy of open disclosure of methodology
and free access to data, we would not have been as successful in advancing this technology as
much as we have over the last two years.
What is RiskMetrics?
RiskMetrics is a set of tools that enable participants in the financial markets to estimate their exposure to market risk under what has been called the Value-at-Risk framework. RiskMetrics has
three basic components:
A set of market risk measurement methodologies outlined in this document.
Data sets of volatility and correlation data used in the computation of market risk.
Software systems developed by J.P.Morgan, subsidiaries of Reuters, and third party vendors
that implement the methodologies described herein.
With the help of this document and the associated line of products, users should be in a position
to estimate market risks in portfolios of foreign exchange, fixed income, equity and commodity
products.
J.P. Morgan and Reuters team up on RiskMetrics
In June 1996, J.P. Morgan signed an agreement with Reuters to cooperate on the building of a new
and more powerful version of RiskMetrics. Since the launch of RiskMetrics in October 1994, we
have received numerous requests to add new products, instruments, and markets to the daily volatility and correlation data sets. We have also perceived the need in the market for a more flexible
VaR data tool than the standard matrices that are currently distributed over the Internet. The new

iv

Preface to the fourth edition

partnership with Reuters, which will be based on the precept that both firms will focus on their
respective strengths, will help us achieve these objectives.
Methodology
J.P. Morgan will continue to develop the RiskMetrics set of VaR methodologies and publish them
in the quarterly RiskMetrics Monitor and in the annual RiskMetricsTechnical Document.
RiskMetrics data sets
Reuters will take over the responsibility for data sourcing as well as production and delivery of the
risk data sets. The current RiskMetrics data sets will continue to be available on the Internet free of
charge and will be further improved as a benchmark tool designed to broaden the understanding of
the principles of market risk measurement.
When J.P. Morgan first launched RiskMetrics in October 1994, the objective was to go for broad
market coverage initially, and follow up with more granularity in terms of the markets and instruments covered. This over time, would reduce the need for proxies and would provide additional
data to measure more accurately the risk associated with non-linear instruments.
The partnership will address these new markets and products and will also introduce a new customizable service, which will be available over the Reuters Web service. The customizable
RiskMetrics approach will give risk managers the ability to scale data to meet the needs of their
individual trading profiles. Its capabilities will range from providing customized covariance matrices needed to run VaR calculations, to supplying data for historical simulation and stress-testing
scenarios.
More details on these plans will be discussed in later editions of the RiskMetrics Monitor.
Systems
Both J.P. Morgan and Reuters, through its Sailfish subsidiary, have developed client-site
RiskMetrics VaR applications. These products, together with the expanding suite of third party
applications will continue to provide RiskMetrics implementations.
What is new in this fourth edition?
In terms of content, the Fourth Edition of the Technical Document incorporates the changes and
refinements to the methodology that were initially outlined in the 19951996 editions of the
RiskMetrics Monitor:
Expanded framework: We have worked extensively on refining the analytical framework
for analyzing options risk without having to perform relatively time consuming simulations
and have outlined the basis for an improved methodology which incorporates better information on the tails of distributions related to financial asset price returns; weve also developed a
data synchronization algorithm to refine our volatility and correlation estimates for products
which do not trade in the same time zone;
New markets: We expanded the daily data sets to include estimated volatilities and correlations of additional foreign exchange, fixed income and equity markets, particularly in South
East Asia and Latin America.
Fine-tuned methodology: We have modified the approach in a number of ways. First, weve
changed our definition of price volatility which is now based on a total return concept; weve
also revised some of the algorithms used in our mapping routines and are in the process of
redefining the techniques used in estimating equity portfolio risk.
RiskMetrics Technical Document
Fourth Edition

Preface to the fourth edition

RiskMetrics products: While we have continued to expand the list of third parties providing
RiskMetrics products and support, this is no longer included with this document. Given the
rapid pace of change in the availability of risk management software products, readers are
advised to consult our Internet web site for the latest available list of products. This list,
which now includes FourFifteen , J.P. Morgans own VaR calculator and report generating
software, continues to grow, attesting to the broad acceptance RiskMetrics has achieved.
New tools to use the RiskMetrics data sets: We have published an Excel add-in function
which enables users to import volatilities and correlations directly into a spreadsheet. This
tool is available from our Internet web site.
The structure of the document has changed only slightly. As before, its size warrants the following
note: One need not read and understand the entire document in order to benefit from RiskMetrics.
The document is organized in parts that address subjects of particular interest to many readers.
Part I:

Risk Measurement Framework

This part is for the general practitioner. It provides a practical framework on how to
think about market risks, how to apply that thinking in practice, and how to interpret the
results. It reviews the different approaches to risk estimation, shows how the calculations work on simple examples and discusses how the results can be used in limit management, performance evaluation, and capital allocation.

Part II:

Statistics of Financial Market Returns

This part requires an understanding and interest in statistical analysis. It reviews the
assumptions behind the statistics used to describe financial market returns and how distributions of future returns can be estimated.

Part III:

Risk Modeling of Financial Instruments

This part is required reading for implementation of a market risk measurement system.
It reviews how positions in any asset class can be described in a standardized fashion
(foreign exchange, interest rates, equities, and commodities). Special attention is given
to derivatives positions. The purpose is to demystify derivatives in order to show that
their market risks can be measured in the same fashion as their underlying.

Part IV:

RiskMetrics Data Sets

This part should be of interest to users of the RiskMetrics data sets. First it describes the
sources of all daily price and rate data. It then discusses the attributes of each volatility
and correlation series in the RiskMetrics data sets. And last, it provides detailed format
descriptions required to decipher the data sets that can be downloaded from public or
commercial sources.

Appendices
This part reviews some of the more technical issues surrounding methodology and regulatory requirements for market risk capital in banks and demonstrates the use of RiskMetrics with the example diskette provided with this document. Finally, Appendix H
shows you how to access the RiskMetrics data sets from the Internet.

vi

Preface to the fourth edition

RiskMetrics examples diskette

This diskette is located inside the back cover. It contains an Excel workbook that
includes some of the examples shown in this document. Such examples are identified by
the icon shown here.
Future plans
We expect to update this Technical Document annually as we adapt our market risk standards to
further improve the techniques and data to meet the changing needs of our clients.
RiskMetrics is a now an integral part of J.P. Morgans Risk Management Services group which
provides advisory services to a wide variety of the firms clients. We continue to welcome any suggestions to enhance the methodology and adapt it further to the needs of the market. All suggestions, requests and inquiries should be directed to the authors of this publication or to your local
RiskMetrics contacts listed on the back cover.
Acknowledgments
The authors would like to thank the numerous individuals who participated in the writing and editing of this document, particularly Chris Finger and Chris Athaide from J.P. Morgans risk management research group, and Elizabeth Frederick and John Matero from our risk advisory practice.
Finally, this document could not have been produced without the contributions of our consulting
editor, Tatiana Kolubayev. We apologize for any omissions to this list.

RiskMetrics Technical Document

Fourth Edition

vii

Table of contents
Part I

Risk Measurement Framework

Chapter 1.
1.1
1.2
1.3

Introduction
An introduction to Value-at-Risk and RiskMetrics
A more advanced approach to Value-at-Risk using RiskMetrics
What RiskMetrics provides

3
6
7
16

Chapter 2.
2.1
2.2
2.3

Historical perspective of VaR

From ALM to VaR
VaR in the framework of modern financial management
Alternative approaches to risk estimation

19
22
24
26

Chapter 3.
3.1
3.2
3.3
3.4

Applying the risk measures

Market risk limits
Calibrating valuation and risk models
Performance evaluation
Regulatory reporting, capital requirement

31
33
34
34
36

Part II

Statistics of Financial Market Returns

Chapter 4.
4.1
4.2
4.3
4.4
4.5
4.6
4.7

Statistical and probability foundations

Definition of financial price changes and returns
Modeling financial prices and returns
Investigating the random-walk model
Summary of our findings
A review of historical observations of return distributions
RiskMetrics model of financial returns: A modified random walk
Summary

Chapter 5.
5.1
5.2
5.3
5.4

Estimation and forecast

Forecasts from implied versus historical information
RiskMetrics forecasting methodology
Estimating the parameters of the RiskMetrics model
Summary and concluding remarks

43
45
49
54
64
64
73
74
75
77
78
90
100

Part III Risk Modeling of Financial Instruments

Chapter 6.
6.1
6.2
6.3
6.4

Market risk methodology

Step 1Identifying exposures and cash flows
Step 2Mapping cash flows onto RiskMetrics vertices
Step 3Computing Value-at-Risk
Examples

105
107
117
121
134

Chapter 7.
7.1
7.2
7.3
7.4

Monte Carlo
Scenario generation
Portfolio valuation
Summary
Comments

149
151
155
157
159

viii

Table of contents

Part IV RiskMetrics Data Sets

Chapter 8.
8.1
8.2
8.3
8.4
8.5

Data and related statistical issues

Constructing RiskMetrics rates and prices
Filling in missing data
The properties of correlation (covariance) matrices and VaR
Rebasing RiskMetrics volatilities and correlations
Nonsynchronous data collection

163
165
170
176
183
184

Chapter 9.
9.1
9.2
9.3
9.4
9.5
9.6

Time series sources

Foreign exchange
Money market rates
Government bond zero rates
Swap rates
Equity indices
Commodities

197
199
199
200
202
203
205

Chapter 10.
10.1
10.2
10.3
10.4
10.5
10.6
10.7

RiskMetrics volatility and correlation files

Availability
File names
Data series naming standards
Format of volatility files
Format of correlation files
Data series order
Underlying price/rate availability

207
209
209
209
211
212
214
214

Part V

Backtesting

Chapter 11.
11.1
11.2
11.3

Performance assessment
Sample portfolio
Assessing the RiskMetrics model
Summary

217
219
220
223

Appendices
Appendix A. Tests of conditional normality

227

Appendix B. Relaxing the assumption of conditional normality

235

Appendix C. Methods for determining the optimal decay factor

243

Appendix D. Assessing the accuracy of the delta-gamma approach

247

Appendix E. Routines to simulate correlated normal random variables

253

Appendix F. BIS regulatory requirements

257

Appendix G. Using the RiskMetrics examples diskette

263

Appendix H. RiskMetrics on the Internet

267

Reference
Glossary of terms

271

Bibliography

275

RiskMetrics Technical Document

Fourth Edition

ix

List of charts
Chart 1.1
Chart 1.2
Chart 1.3
Chart 1.4
Chart 1.5
Chart 1.6
Chart 1.7
Chart 1.8
Chart 1.9
Chart 2.1
Chart 2.2
Chart 2.3
Chart 2.4
Chart 3.1
Chart 3.2
Chart 3.3
Chart 3.4
Chart 3.5
Chart 4.1
Chart 4.2
Chart 4.3
Chart 4.4
Chart 4.5
Chart 4.6
Chart 4.7
Chart 4.8
Chart 4.9
Chart 4.10
Chart 4.11
Chart 4.12
Chart 4.13
Chart 4.14
Chart 4.15
Chart 4.16
Chart 4.17
Chart 4.18
Chart 4.19
Chart 4.20
Chart 4.21
Chart 5.1
Chart 5.2
Chart 5.3
Chart 5.4
Chart 5.5
Chart 5.6
Chart 5.7
Chart 5.8
Chart 5.9
Chart 5.10
Chart 5.11
Chart 5.12
Chart 6.1

VaR statistics
6
Simulated portfolio changes
9
Actual cash flows
9
Mapping actual cash flows onto RiskMetrics vertices
10
Value of put option on USD/DEM
14
Histogram and scattergram of rate distributions
15
Valuation of instruments in sample portfolio
15
Representation of VaR
16
Components of RiskMetrics
17
Asset liability management
22
Value-at-Risk management in trading
23
Comparing ALM to VaR management
24
Two steps beyond accounting
25
Hierarchical VaR limit structure
33
Ex post validation of risk models: DEaR vs. actual daily P&L
34
Performance evaluation triangle
35
Example: comparison of cumulative trading revenues
35
Example: applying the evaluation triangle
36
Absolute price change and log price change in U.S. 30-year government bond
47
Simulated stationary/mean-reverting time series
52
Simulated nonstationary time series
53
Observed stationary time series
53
Observed nonstationary time series
54
USD/DEM returns
55
USD/FRF returns
55
Sample autocorrelation coefficients for USD/DEM foreign exchange returns
57
Sample autocorrelation coefficients for USD S&P 500 returns
58
USD/DEM returns squared
60
S&P 500 returns squared
60
Sample autocorrelation coefficients of USD/DEM squared returns
61
Sample autocorrelation coefficients of S&P 500 squared returns
61
Cross product of USD/DEM and USD/FRF returns
63
Correlogram of the cross product of USD/DEM and USD/FRF returns
63
Leptokurtotic vs. normal distribution
65
Normal distribution with different means and variances
67
Selected percentile of standard normal distribution
69
One-tailed confidence interval
70
Two-tailed confidence interval
71
Lognormal probability density function
73
DEM/GBP exchange rate
79
80
Log price changes in GBP/DEM and VaR estimates (1.65)
NLG/DEM exchange rate and volatility
87
88
S&P 500 returns and VaR estimates (1.65)
GARCH(1,1)-normal and EWMA estimators
90
USD/DEM foreign exchange
92
Tolerance level and decay factor
94
Relationship between historical observations and decay factor
95
95
Exponential weights for T = 100
One-day volatility forecasts on USD/DEM returns
96
One-day correlation forecasts for returns on USD/DEM FX rate and on S&P500 96
Simulated returns from RiskMetrics model
101
French franc 10-year benchmark maps
109

List of charts

Chart 6.2
Chart 6.3
Chart 6.4
Chart 6.5
Chart 6.6
Chart 6.7
Chart 6.8
Chart 6.9
Chart 6.10
Chart 6.11
Chart 6.12
Chart 6.13
Chart 6.14
Chart 6.15
Chart 6.16
Chart 6.17
Chart 6.18
Chart 6.19
Chart 7.1
Chart 7.2
Chart 7.3
Chart 7.4
Chart 7.5
Chart 8.1
Chart 8.2
Chart 8.3
Chart 8.4
Chart 8.5
Chart 8.6
Chart 8.7
Chart 8.8
Chart 9.1
Chart 11.1
Chart 11.2
Chart 11.3
Chart 11.4
Chart A.1
Chart A.2
Chart A.3
Chart B.1
Chart B.2
Chart B.3
Chart D.1
Chart D.2

Cash flow representation of a simple bond

Cash flow representation of a FRN
Estimated cash flows of a FRN
Cash flow representation of simple interest rate swap
Cash flow representation of forward starting swap
Cash flows of the floating payments in a forward starting swap
Cash flow representation of FRA
Replicating cash flows of 3-month vs. 6-month FRA
Cash flow representation of 3-month Eurodollar future
Replicating cash flows of a Eurodollar futures contract
FX forward to buy Deutsche marks with US dollars in 6 months
Replicating cash flows of an FX forward
Actual cash flows of currency swap
RiskMetrics cash flow mapping
Linear and nonlinear payoff functions
VaR horizon and maturity of money market deposit
Long and short option positions
DEM 3-year swaps in Q1-94
Frequency distributions for and
Frequency distribution for DEM bond price
Frequency distribution for USD/DEM exchange rate
Value of put option on USD/DEM
Distribution of portfolio returns
Constant maturity future: price calculation
Graphical representation
Number of variables used in EM and parameters required
Correlation forecasts vs. return interval
Time chart
10-year Australia/US government bond zero correlation
Adjusting 10-year USD/AUD bond zero correlation
10-year Japan/US government bond zero correlation
Volatility estimates: daily horizon
One-day Profit/Loss and VaR estimates
Histogram of standardized returns
Standardized lower-tail returns
Standardized upper-tail returns
Standard normal distribution and histogram of returns on USD/DEM
Quantile-quantile plot of USD/DEM
Quantile-quantile plot of 3-month sterling
Tails of normal mixture densities
GED distribution
Left tail of GED () distribution
Delta vs. time to expiration and underlying price
Gamma vs. time to expiration and underlying price

RiskMetrics Technical Document

Fourth Edition

109
110
111
111
112
113
113
114
114
114
115
115
116
118
123
128
131
141
153
154
154
157
158
170
175
176
185
188
190
194
195
202
219
221
222
222
227
232
234
238
239
240
248
249

xi

List of tables
Table 2.1
Table 3.1
Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 4.5
Table 4.6
Table 4.7
Table 5.1
Table 5.2
Table 5.3
Table 5.4
Table 5.5
Table 5.6
Table 5.7
Table 5.8
Table 5.9
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7
Table 6.8
Table 6.9
Table 6.10
Table 6.11
Table 6.12
Table 6.13
Table 6.14
Table 6.15
Table 6.16
Table 6.17
Table 6.18
Table 6.19
Table 6.20
Table 7.1
Table 7.2
Table 7.3
Table 8.1
Table 8.2
Table 8.3
Table 8.4
Table 8.5
Table 8.6
Table 8.7
Table 8.7
Table 8.8
Table 8.9
Table 8.10

Two discriminating factors to review VaR models

Comparing the Basel Committee proposal with RiskMetrics
Absolute, relative and log price changes
Return aggregation
Box-Ljung test statistic
Box-Ljung statistics
Box-Ljung statistics on squared log price changes (cv = 25)
Model classes
VaR statistics based on RiskMetrics and BIS/Basel requirements
Volatility estimators
Calculating equally and exponentially weighted volatility
Applying the recursive exponential weighting scheme to compute volatility
Covariance estimators
Recursive covariance and correlation predictor
Mean, standard deviation and correlation calculations
The number of historical observations used by the EWMA model
Optimal decay factors based on volatility forecasts
Summary of RiskMetrics volatility and correlation forecasts
Data provided in the daily RiskMetrics data set
Data calculated from the daily RiskMetrics data set
Relationship between instrument and underlying price/rate
Statistical features of an option and its underlying return
RiskMetrics data for 27, March 1995
RiskMetrics map of single cash flow
RiskMetrics map for multiple cash flows
Mapping a 6x12 short FRF FRA at inception
Mapping a 6x12 short FRF FRA held for one month
Structured note specification
Actual cash flows of a structured note
VaR calculation of structured note
VaR calculation on structured note
Cash flow mapping and VaR of interest-rate swap
VaR on foreign exchange forward
Market data and RiskMetrics estimates as of trade date July 1, 1994
Cash flow mapping and VaR of commodity futures contract
Portfolio specification
Portfolio statistics
Value-at-Risk estimates (USD)
Monte Carlo scenarios
Monte Carlo scenariosvaluation of option
Value-at-Risk for example portfolio
Construction of rolling nearby futures prices for Light Sweet Crude (WTI)
Price calculation for 1-month CMF NY Harbor #2 Heating Oil
Belgian franc 10-year zero coupon rate
Singular values for USD yield curve data matrix
Singular values for equity indices returns
Correlations of daily percentage changes with USD 10-year
Schedule of data collection
Schedule of data collection
RiskMetrics closing prices
Sample statistics on RiskMetrics daily covariance forecasts
RiskMetrics daily covariance forecasts

29
39
46
49
58
59
62
66
71
78
81
82
83
84
91
94
99
100
121
121
123
130
134
134
135
137
137
139
139
140
140
142
143
145
145
147
148
148
155
156
158
168
169
175
182
182
184
186
187
191
191
192

xii

List of tables

Table 8.11
Table 8.12
Table 8.13
Table 9.1
Table 9.2
Table 9.3
Table 9.4
Table 9.5
Table 9.6
Table 9.7
Table 9.8
Table 10.1
Table 10.2
Table 10.3
Table 10.4
Table 10.5
Table 10.6
Table 10.7
Table 11.1
Table 11.2
Table A.1
Table A.2
Table B.1
Table B.2
Table B.3
Table D.1
Table D.2
Table D.3

Relationship between lagged returns and applied weights

Original and adjusted correlation forecasts
Correlations between US and foreign instruments
Foreign exchange
Money market rates: sources and term structures
Government bond zero rates: sources and term structures
Swap zero rates: sources and term structures
Equity indices: sources
Commodities: sources and term structures
Energy maturities
Base metal maturities
RiskMetrics file names
Currency and commodity identifiers
Maturity and asset class identifiers
Sample volatility file
Data columns and format in volatility files
Sample correlation file
Data columns and format in correlation files
Realized percentages of VaR violations
Realized tail return averages
Sample mean and standard deviation estimates for USD/DEM FX
Testing for univariate conditional normality
Parameter estimates for the South African rand
Sample statistics on standardized returns
VaR statistics (in %) for the 1st and 99th percentiles
Parameters used in option valuation
MAPE (%) for call, 1-day forecast horizon
ME (%) for call, 1-day forecast horizons

RiskMetrics Technical Document

Fourth Edition

193
193
196
199
200
201
203
204
205
205
206
209
210
210
211
212
213
213
220
221
228
230
240
241
242
249
251
251

Part I
Risk Measurement Framework

RiskMetrics Technical Document

Fourth Edition

Chapter 1.

Introduction
1.1 An introduction to Value-at-Risk and RiskMetrics
1.2 A more advanced approach to Value-at-Risk using RiskMetrics
1.2.1
Using RiskMetrics to compute VaR on a portfolio of cash flows
1.2.2
Measuring the risk of nonlinear positions
1.3 What RiskMetrics provides
1.3.1
An overview
1.3.2
Detailed specification

Part I: Risk Measurement Framework

6
7
9
11
16
16
18

RiskMetrics Technical Document

Fourth Edition

Chapter 1.

Introduction

Jacques Longerstaey
Morgan Guaranty Trust Company
Risk Management Advisory
(1-212) 648-4936
[email protected]

This chapter serves as an introduction to the RiskMetrics product. RiskMetrics is a set of methodologies and data for measuring market risk. By market risk, we mean the potential for changes in
value of a position resulting from changes in market prices.
We define risk as the degree of uncertainty of future net returns. This uncertainty takes many
forms, which is why most participants in the financial markets are subject to a variety of risks. A
common classification of risks is based on the source of the underlying uncertainty:
Credit risk estimates the potential loss because of the inability of a counterparty to meet its
obligations.
Operational risk results from errors that can be made in instructing payments or settling transactions.
Liquidity risk is reflected in the inability of a firm to fund its illiquid assets.
Market risk, the subject of the methodology described in this document, involves the uncertainty of future earnings resulting from changes in market conditions, (e.g., prices of assets,
interest rates). Over the last few years measures of market risk have become synonymous
with the term Value-at-Risk.
RiskMetrics has three basic components:
The first is a set of methodologies outlining how risk managers can compute Value-at-Risk on
a portfolio of financial instruments. These methodologies are explained in this
Technical Document, which is an annual publication, and in the RiskMetrics Monitor, the
quarterly update to the Technical Document.
The second is data that we distribute to enable market participants to carry out the methodologies set forth in this document.
The third is Value-at-Risk calculation and reporting software designed by J.P. Morgan,
Reuters, and third party developers. These systems apply the methodologies set forth in this
document and will not be discussed in this publication.
This chapter is organized as follows:
Section 1.1 presents the definition of Value-at-Risk (VaR) and some simple examples of how
RiskMetrics offers the inputs necessary to compute VaR. The purpose of this section is to
offer a basic approach to VaR calculations.
Section 1.2 describes more detailed examples of VaR calculations for a more thorough understanding of how RiskMetrics and VaR calculations fit together. In Section 1.2.2 we provide an
example of how to compute VaR on a portfolio containing options (nonlinear risk) using two
different methodologies.
Section 1.3 presents the contents of RiskMetrics at both the general and detailed level. This
section provides a step-by-step analysis of the production of RiskMetrics volatility and correlation files as well as the methods that are necessary to compute VaR. For easy reference we
provide section numbers within each step so that interested readers can learn more about that
particular subject.

Part I: Risk Measurement Framework

Chapter 1. Introduction

Reading this chapter requires a basic understanding of statistics. For assistance, readers can refer
to the glossary at the end of this document.
1.1 An introduction to Value-at-Risk and RiskMetrics
Value-at-Risk is a measure of the maximum potential change in value of a portfolio of financial
instruments with a given probability over a pre-set horizon. VaR answers the question: how much
can I lose with x% probability over a given time horizon. For example, if you think that there is a
95% chance that the DEM/USD exchange rate will not fall by more than 1% of its current value
over the next day, you can calculate the maximum potential loss on, say, a USD 100 million
DEM/USD position by using the methodology and data provided by RiskMetrics. The following
examples describe how to compute VaR using standard deviations and correlations of financial
returns (provided by RiskMetrics) under the assumption that these returns are normally distributed. (RiskMetrics provides alternative methodological choices to address the inacurracies resulting from this simplifying assumption).
Example 1: You are a USD-based corporation and hold a DEM 140 million FX position. What
is your VaR over a 1-day horizon given that there is a 5% chance that the realized loss will be
greater than what VaR projected? The choice of the 5% probability is discretionary and differs
across institutions using the VaR framework.
What is your exposure?

The first step in the calculation is to compute your exposure

to market risk (i.e., mark-to-market your position). As a USDbased investor, your exposure is equal to the market value of
the position in your base currency. If the foreign exchange
rate is 1.40 DEM/USD, the market value of the position is
USD 100 million.

What is your risk?

Moving from exposure to risk requires an estimate of how

much the exchange rate can potentially move. The standard
deviation of the return on the DEM/USD exchange rate, measured historically can provide an indication of the size of rate
movements. In this example, we calculated the DEM/USD
daily standard deviation to be 0.565%. Now, under the standard RiskMetrics assumption that standardized returns
( ( r t t ) on DEM/USD are normally distributed given the
value of this standard deviation, VaR is given by 1.65 times
the standard deviation (that is, 1.65) or 0.932% (see
Chart 1.1). This means that the DEM/USD exchange rate is
not expected to drop more than 0.932%, 95% of the time.
RiskMetrics provides users with the VaR statistics 1.65.

Chart 1.1
VaR statistics
No. of observations

In USD, the VaR of the position1 is equal to the market value

of the position times the estimated volatility or:

5%
rt/t

FX Risk: $100 million 0.932% = $932,000

What this number means is that 95% of the time, you will not
lose more than $932,000 over the next 24 hours.

This is a simple approximation.

RiskMetrics Technical Document

Fourth Edition

Sec. 1.2 A more advanced approach to Value-at-Risk using RiskMetrics

Example 2: Lets complicate matters somewhat. You are a USD-based corporation and hold a
DEM 140 million position in the 10-year German government bond. What is your VaR over a
1-day horizon period, again, given that there is a 5% chance of understating the realized loss?
What is your exposure?

The only difference versus the previous example is that you

now have both interest rate risk on the bond and FX risk resulting from the DEM exposure. The exposure is still USD 100
million but it is now at risk to two market risk factors.

What is your risk?

If you use an estimate of 10-year German bond standard deviation of 0.605%, you can calculate:
Interest rate risk: $100 million 1.65 0.605% = $999,000
FX Risk: $100 million 1.65 0.565% = $932,000
Now, the total risk of the bond is not simply the sum of the
interest rate and FX risk because the correlation2 between the
return on the DEM/USD exchange rate the return on the 10year German bond is relevant. In this case, we estimated the
correlation between the returns on the DEM/USD exchange
rate and the 10-year German government bond to be 0.27.
Using a formula common in standard portfolio theory, the total
risk of the position is given by:

[1.1]

VaR =

2
Interest rate

+ FX + ( 2 Interest rate, FX Interest rate FX )

VaR = ( 0.999 ) 2 + ( 0.932 ) 2 + ( 2 0.27 0.999 0.932 )
= $ 1.168 million
To compute VaR in this example, RiskMetrics provides
users with the VaR of interest rate component
(i.e., 1.65 0.605), the VaR of the foreign exchange position
(i.e., 1.65 0.565) and the correlation between the two
return series, 0.27.

1.2 A more advanced approach to Value-at-Risk using RiskMetrics

Value-at-Risk is a number that represents the potential change in a portfolios future value. How
this change is defined depends on (1) the horizon over which the portfolios change in value is
measured and (2) the degree of confidence chosen by the risk manager.
VaR calculations can be performed without using standard deviation or correlation forecasts.
These are simply one set of inputs that can be used to calculate VaR, and that RiskMetrics provides for that purpose. The principal reason for preferring to work with standard deviations (volatility) is the strong evidence that the volatility of financial returns is predictable. Therefore, if
volatility is predictable, it makes sense to make forecasts of it to predict future values of the return
distribution.

Correlation is a measure of how two series move together. For example, a correlation of 1 implies that two series
move perfectly together in the same direction.

Part I: Risk Measurement Framework

Chapter 1. Introduction

Suppose we want to compute the Value-at-Risk of a portfolio over a 1-day horizon with a 5%
chance that the actual loss in the portfolios value is greater than the VaR estimate. The Value-atRisk calculation consists of the following steps.
1.

Mark-to-market the current portfolio. Denote this value by V 0 .

2.

Define the future value of the portfolio, V 1 , as V 1 = V 0 e where3 r represents the return
on the portfolio over the horizon. For a 1-day horizon, this step is unnecessary as
RiskMetrics assumes a 0 return.

3.

Make a forecast of the 1-day return on the portfolio and denote this value by r , such that
there is a 5% chance that the actual return will be less than r . Alternatively expressed,

Probability ( r < r ) = 5%.

4.

r
Define the portfolios future worst case value V 1 , as V 1 = V 0 e . The Value-at-Risk estimate is simply V V 1 .
0

r
Notice that the VaR estimate can be written as V 0 1 e . In the case that r is sufficiently
r
small, e 1 + r so that VaR is approximately equal toV 0 r . is approximately equal to V 0 r .
The purpose of a risk measurement system such as RiskMetrics is to offer a means to compute r .
Within this more general framework we use a simple example to demonstrate how the RiskMetrics
methodologies and data enable users to compute VaR. Assume the forecast horizon over which
VaR is measured is one day and the level of confidence in the forecast to 5%. Following the
steps outlined above, the calculation would proceed as follows:
1.

Consider a portfolio whose current marked-to-market value, V 0 , is USD 500 million.

2.

To carry out the VaR calculation we require 1-day forecasts of the mean 1 0 . Within the
RiskMetrics framework, we assume that the mean return over a 1-day horizon period is
equal to 0.

3.

We also need the standard deviation, 1 0 , of the returns in this portfolio. Assuming that the
return on this portfolio is distributed conditionally normal, r = 1.65 1 0 + 1 0 . The
RiskMetrics data set provides the term 1.65 1 0 . Hence, setting 1 0 = 0 and
1 0 = 0.0321 , we get V 1 = USD 474.2 million .4

4.

This yields a Value-at-Risk of USD 25.8 million (given by V 0 V 1 ).

The histogram in Chart 1.2 presents future changes in value of the portfolio. VaR reduces risk to
just one number, i.e., a loss associated with a given probability. It is often useful for risk managers
to focus on the total distribution of potential gains and losses and we will discuss why this is so
later in this document. (See Section 6.3).

Where e is approximately 2.27183

This number is computed from e

1.65

RiskMetrics Technical Document

Fourth Edition

V0

Sec. 1.2 A more advanced approach to Value-at-Risk using RiskMetrics

Chart 1.2
Simulated portfolio changes
Probability
0.10
0.09
0.08

95% confidence:

0.07
$25.8 million

0.06
0.05
0.04
0.03
0.02
0.01
0.00

-48 -40 -32 -24 -16 -8

16

24

32

40

48

P/L ($million)

1.2.1 Using RiskMetrics to compute VaR on a portfolio of cash flows

Calculating VaR usually involves more steps than the basic ones outlined in the examples above.
Even before calculating VaR, you need to estimate to which risk factors a particular portfolio is
exposed. The preferred methodology for doing this is to decompose financial instruments into their
basic cash flow components. The RiskMetrics methodology and data allow users to compute the
VaR on portfolios consisting of a variety of cash flows. We use a simple example (a portfolio consisting of three cash flows) to demonstrate how to compute VaR.
Step 1. Each financial position in a portfolio is expressed as one or more cash flows that are
marked-to-market at current market rates. For example, consider an instrument that gives
rise to three USD 100 cash flows each occurring in 1, 4, and 7 months time as shown in
Chart 1.3.
Chart 1.3
Actual cash flows
100

100

100

1m

4m

7m

Principal flows

Step 2. When necessary, the actual cash flows are converted to RiskMetrics cash flows by mapping (redistributing) them onto a standard grid of maturity vertices, known as RiskMetrics
vertices, which are fixed at the following intervals:
1m

3m

6m

12m 2yr

3yr

4yr

5yr

7yr

9yr

10yr 15yr 20yr 30yr

The purpose of the mapping is to standardize the cash flow intervals of the instrument such
that we can use the volatilities and correlations that are routinely computed for the given
vertices in the RiskMetrics data sets. (It would be impossible to provide volatility and correlation estimates on every possible maturity so RiskMetrics provides a mapping method-

Part I: Risk Measurement Framework

10

Chapter 1. Introduction

ology which distributes cash flows to a workable set of standard maturities). The
methodology for mapping cash flows is detailed in Chapter 6.
To map the cash flows, we use the RiskMetrics vertices closest to the actual vertices and
redistribute the actual cash flows as shown in Chart 1.4.
Chart 1.4
Mapping actual cash flows onto RiskMetrics vertices
100

100

100

1m

4m

7m

100

60

1m

70

40

Actual cashflows
30

3m

6m

12m

100

60

110

30

1m

3m

6m

12m

Cashflow mapping

RiskMetrics cashflows

The RiskMetrics cash flow map is used to work backwards to calculate the return for each
of the actual cash flows from the cash flow at the associated RiskMetrics vertex, or vertices.
For each actual cash flow, an analytical expression is used to express the relative change in
value of the actual cash flow in terms of an underlying return on a particular instrument.
Continuing with Chart 1.4, we can write the return on the actual 4-month cash flow in
terms of the combined returns on the 3-month (60%) and 6-month (40%) RiskMetrics cash
flows:
[1.2]

r 4m = 0.60r 3m + 0.40r 6m

where
r 4m = return on the actual 4-month cash flow
r 3m = return on the 3-month RiskMetrics cash flow
r 6m = return on the 6-month RiskMetrics cash flow
Similarly, the return on the 7-month cash flow can be written as
[1.3]

r 7m = 0.70r 6m + 0.30r 12m

Note that the return on the actual 1-month cash flow is equal to the return on the 1-month
instrument.
Step 3. VaR is calculated at the 5th percentile of the distribution of portfolio return, and for a specified time horizon. In the example above, the distribution of the portfolio return, r p , is
written as:
[1.4]

r p = 0.33r 1m + 0.20r 3m + 0.37r 6m + 0.10r 12m

RiskMetrics Technical Document

Fourth Edition

Sec. 1.2 A more advanced approach to Value-at-Risk using RiskMetrics

11

where, for example the portfolio weight 0.33 is the result of 100 divided by the total portfolio value 300.
Now, to compute VaR at the 95th percent confidence level we need the fifth percentile of
the portfolio return distribution. Under the assumption that r p is distributed conditionally
normal, the fifth percentile is 1.65 p where p is the standard deviation of the portfolio
return distribution. Applying Eq. [1.1] to a portfolio containing more than two instruments
requires using simple matrix algebra. We can thus express this VaR calculation as follows:
[1.5]

VaR =

V RV

where V is a vector of VaR estimates per instrument,

V = [ ( 0.33 1.65 1m ) , ( 0.20 1.65 3m ) , ( 0.37 1.65 6m ) , ( 0.10 1.65 12m ) ] ,
and R is the correlation matrix
1
[1.6]

R =

1m, 3m

3m, 1m 6m, 1m 12m, 1m

1

1m, 6m 3m, 6m

6m, 3m 12m, 3m
1

1m, 12m 3m, 12m 6m, 12m

12m, 6m
1

where, for example, 1m, 3m is the correlation estimate between 1-month and 3-month
returns.
Note that RiskMetrics provides the vector of information
V = [ ( 1.65 1m ) , ( 1.65 3m ) , ( 1.65 6m ) , ( 1.65 12m ) ]
as well as the correlation matrix R. What the user has to provide are the actual portfolio weights.
1.2.2 Measuring the risk of nonlinear positions
When the relationship between position value and market rates is nonlinear, then we cannot estimate changes in value by multiplying estimated changes in rates by sensitivity of the position
to changing rates; the latter is not constant (i.e., the definition of a nonlinear position). In our previous examples, we could easily estimate the risk of a fixed income or foreign exchange product
by assuming a linear relationship between the value of an instrument and the value of its underlying. This is not a reasonable assumption when dealing with nonlinear products such as options.
RiskMetrics offers two methodologies, an analytical approximation and a structured Monte
Carlo simulation to compute the VaR of nonlinear positions:
1.

The first method approximates the nonlinear relationship via a mathematical expression
that relates the return on the position to the return on the underlying rates. This is done by
using what is known as a Taylor series expansion.
This approach no longer necessarily assumes that the change in value of the instrument is
approximated by its delta alone (the first derivative of the options value with respect to the
underlying variable) but that a second order term using the options gamma (the second
derivative of the options value with respect to the underlying price) must be introduced to

Part I: Risk Measurement Framework

12

Chapter 1. Introduction

measure the curvature of changes in value around the current value. In practice, other
greeks such as vega (volatility), rho (interest rate) and theta (time to maturity) can also be
used to improve the accuracy of the approximation. In Section 1.2.2.1, we present two
types of analytical methods for computing VaRthe delta and delta-gamma approximation.
2.

The second alternative, structured Monte Carlo simulation, involves creating a large number of possible rate scenarios and revaluing the instrument under each of these scenarios.
VaR is then defined as the 5th percentile of the distribution of value changes. Due to the
required revaluations, this approach is computationally more intensive than the first
approach.

The two methods differ not in terms of how market movements are forecast (since both use the
RiskMetrics volatility and correlation estimates) but in how the value of portfolios changes as a
result of market movements. The analytical approach approximates changes in value, while the
structured Monte Carlo fully revalues portfolios under various scenarios.
Let us illustrate these two methods using a practical example. We will consider throughout this
section a portfolio comprised of two assets:
Asset 1: a future cash flow stream of DEM 1 million to be received in one years time. The current 1-year DEM rate is 10% so the current market value of the instrument is DEM 909,091.
Asset 2: an at-the-money (ATM) DEM put/USD call option with contract size of
DEM 1 million and expiration date one month in the future. The premium of the option is 0.0105
and the spot exchange rate at which the contract was concluded is 1.538 DEM/USD. We assume
the implied volatility at which the option is priced is 14%.
The value of this portfolio depends on the USD/DEM exchange rate and the one-year DEM bond
price. Technically, the value of the option also changes with USD interest rates and the implied
volatility, but we will not consider these effects. Our risk horizon for the example will be five days.
We take as the daily volatilities of these two assets FX = 0.42% and B = 0.08% and as the
correlation between the two = 0.17 .
Both alternatives will focus on price risk exclusively and therefore ignore the risk associated with
volatility (vega), interest rate (rho) and time decay (theta risk).
1.2.2.1 Analytical method
There are various ways to analytically approximate nonlinear VaR. This section reviews the two
alternatives which we discussed previously.
Delta approximation
The standard VaR approach can be used to come up with first order approximations of portfolios
that contain options. (This is essentially the same simplification that fixed income traders use when
they focus exclusively on the duration of their portfolio). The simplest such approximation is to
estimate changes in the option value via a linear model, which is commonly known as the delta
approximation. Delta is the first derivative of the option price with respect to the spot exchange
rate. The value of for the option in this example is 0.4919.
In the analytical method, we must first write down the return on the portfolio whose VaR we are
trying to calculate. The return on this portfolio consisting of a cash flow in one year and a put on
the DEM/call on the USD is written as follows:
[1.7]

r p =r 1 y + r DEM + r DEM
-------------USD

-------------USD

RiskMetrics Technical Document

Fourth Edition

Sec. 1.2 A more advanced approach to Value-at-Risk using RiskMetrics

13

where
r 1 p = the price return on the 1-year German interest rates
r DEM = the return on the DEM/USD exchange rate
-------------USD

= the delta of the option

Under the assumption that the portfolio return is normally distributed, VaR at the 95% confidence
level is given by
[1.8]

2 2

VaR = 1.65 1 y + ( 1 + ) DEM + 2 ( 1 + )

-------------USD

DEM 1 y DEM
-------------1 y, -------------USD
USD

Using our volatilities and correlations forecasts for DEM/USD and the 1-year DEM rate (scaled up
to the weekly horizon using the square root of time rule), the weekly VaR for the portfolio using
the delta equivalent approach can be approximated by:

1-yr DEM cash flow

FX position - FX hedge

VaR(1w)
$1,745
$4,654

Market value in USD

$591,086
$300,331

Diversified VaR
$4,684
Delta-gamma approximation
The delta approximation is reasonably accurate when the exchange rate does not change significantly, but less so in the more extreme cases. This is because the delta is a linear approximation of
a non linear relationship between the value of the exchange rate and the price of the option as
shown in Chart 1.5. We may be able to improve this approximation by including the gamma term,
which accounts for nonlinear (i.e. squared returns) effects of changes in the spot rate (this attempts
to replicate the convex option price to FX rate relationship as shown in Chart 1.5). The expression
for the portfolio return is now
[1.9]

r p =r 1 y + r DEM + r DEM + 0.5 P DEM r DEM

---------------------------------------- --------------
USD

USD

USD

USD

where
P DEM = the value of the DEM/USD exchange rate when the VaR forecast is made
-------------USD

= the gamma of the option.

In this example, = DEM/USD 15.14.
Now, the gamma term (the fourth term in Eq. [1.9]) introduces skewness into the distribution of
r P (i.e., the distribution is no longer symmetrical around its mean). Therefore, since this violates
one of the assumptions of normality (symmetry) we can no longer calculate the 95th percentile
VaR as 1.65 times the standard deviation of r p . Instead we must find the appropriate multiple (the
counterpart to 1.65) that incorporates the skewness effect. We compute the 5th percentile of r p s
distribution (Eq. [1.9]) by computing its first four moments, i.e., r p s mean, variance, skewness
and kurtosis. We then find distribution whose first four moments match those of r p s. (See
Section 6.3 for details.)

Part I: Risk Measurement Framework

14

Chapter 1. Introduction

Applying this methodology to this approach we find the VaR for this portfolio to be USD 3,708.
Note that in this example, incorporating gamma reduces VaR relative to the delta only approximation (from USD 5006 to USD 3708).
Chart 1.5
Value of put option on USD/DEM
strike = 0.65 USD/DEM. Value in USD/DEM.
Option value
0.06
0.05

Full valuation

0.04
0.03
0.02
0.01

Delta + gamma

0
-0.01

Delta

-0.02
0.60 0.61 0.62 0.63 0.64 0.65 0.66 0.67 0.68 0.69 0.70
USD/DEM exchange rate

1.2.2.2 Structured Monte-Carlo Simulation

Given the limitations of analytical VaR for portfolios whose P/L distributions may not be symmetrical let alone normally distributed, another possible route is to use a model which instead of estimating changes in value by the product of a rate change () and a sensitivity (, ), focuses on
revaluing positions at changed rate levels. This approach is based on a full valuation precept where
all instruments are marked to market under a large number of scenarios driven by the volatility and
correlation estimates.
The Monte Carlo methodology consists of three major steps:
1.

Scenario generation Using the volatility and correlation estimates for the underlying
assets in our portfolio, we produce a large number of future price scenarios in accordance
with the lognormal models described previously. The methodology for generating scenarios
from volatility and correlation estimates is described in Appendix E.

2.

Portfolio valuation For each scenario, we compute a portfolio value.

3.

Summary We report the results of the simulation, either as a portfolio distribution or as

a particular risk measure.

Using our volatility and correlation estimates, we can apply our simulation technique to our example portfolio. We can generate a large number of scenarios (1000 in this example case) of DEM
1-year and DEM/USD exchange rates at the 1-week horizon. Chart 1.6 shows the actual distributions for both instruments as well as the scattergram indicating the degree of correlation (0.17)
between the two rate series.

RiskMetrics Technical Document

Fourth Edition

Sec. 1.2 A more advanced approach to Value-at-Risk using RiskMetrics

15

Chart 1.6
Histogram and scattergram of rate distributions
2-yr DEM rate and DEM/USD rate
Frequency

Frequency

120

120

100

100

J J JJ J JJ J JJ
J
J J JJJ JJ JJJJJJJ JJ JJJ J
JJ J JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ JJJJJJJJJJJJJJJJ J
J JJJJ JJJJJJJJJJJJJJJJJJJJJJJJ JJJ J
JJ JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ J
J JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
JJJ J J
J
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ J JJ
J J J J JJ JJ
J JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ J
J
JJJ J J JJ J J
J JJJJJ JJJJJJ JJJJJJJJJJJJJJJJ JJJJJJJ J
J J J JJJJJ JJ
J
JJ
J

80
60
40
20
0
9.3% 9.5% 9.7% 10.0%10.2%10.4%10.6%

80
60
40
20
0
1.49 1.50 1.52 1.53 1.55 1.56 1.58

Yields

P/L

With the set of interest and foreign exchange rates obtained under simulation, we can revalue both
of the instruments in our portfolio. Their respective payouts are shown in Chart 1.7.
Chart 1.7
Valuation of instruments in sample portfolio
Value of the cash flow stream

Value of the FX option

Cash flow value - USD thousands

Option value - USD thousands

595.0

25
20

592.5

15
590.0
10
587.5

585.0
9.30

9.55

9.80

10.05

10.30

10.55

Yield

0
1.48

1.5

1.52

1.54

1.56

1.58

1.6

DEM/USD

The final task is to analyze the distribution of values and select the VaR using the appropriate percentile. Chart 1.8 shows the value of the components of the portfolio at the end of the horizon
period.

Part I: Risk Measurement Framework

16

Chapter 1. Introduction

Chart 1.8
Representation of VaR
Histogram of portfolio values
Frequency

Percent
100

100
90

Current value = USD 601,388

95% percentile = USD 596,829
VaR = USD 4,599

80
70

90
80
70

60

60

50

50

40

40

30

30

20

20

10

10

0
595

598

601

604

607

610

614

595

Portfolio value

598

601

604

607

610

614

Portfolio value

The charts above provide a visual indication as to why the delta approximation is usually not suitable for portfolios that contain options. The distribution of returns in portfolios that include
options is typically skewed. The standard delta equivalent VaR approach expects symmetry around
the mean and applies a basic normal distribution approach (i.e., the 95% percentile equates to a
1.65 standard deviation move). In this case, the lack of symmetry in the distribution does not allow
us to apply the normal approximation. Furthermore, the distributions skewness results in a VaR
number that is basically position dependent (i.e., the risk is different whether you are long or short
the option).
1.3 What RiskMetrics provides
As discussed previously, RiskMetrics has three basic components which are detailed below.
1.3.1 An overview
With RiskMetrics J.P. Morgan and Reuters provide
1.

A set of methodologies for statistical market risk measures that are based on, but differ significantly from, the methodology developed and used within J.P. Morgan. This approach
was developed so as to enable other financial institutions, corporate treasuries, and investors to estimate their market risks in a consistent and reasonable fashion. Methodology
defines how positions are to be mapped and how potential market movements are estimated
and is detailed in the following chapters.

2.

Daily recomputed data sets which are comprehensive sets of consistently estimated instrument level VaRs (i.e., 1.65 standard deviations) and correlations across a large number of
asset classes and instruments. We currently distribute three different data sets over the
Internet: one for short term trading risks, the second for intermediate term investment risks
and the third for regulatory reporting. These are made available to the market free of
charge.
In the near future, a more customizable version of RiskMetrics where users will be able to
create covariance matrices from a large underlying database according to various numerical
methods will be made available over the Reuters Web. This product will not replace the

RiskMetrics Technical Document

Fourth Edition

Sec. 1.3 What RiskMetrics provides

17

data sets available over the Internet but will provide subscribers to the Reuters services
with a more flexible tool.
The four basic classes of instruments that RiskMetrics methodology and data sets cover are
represented as follows:
Fixed income instruments are represented by combinations of amounts of cash flows in
a given currency at specified dates in the future. RiskMetrics applies a fixed number of
dates (14 vertices) and two types of credit standings: government and non-government. The data sets associated with fixed income are zero coupon instrument VaR statistics, i.e., 1.65, and correlations for both government and swap yield curves.
Foreign exchange transactions are represented by an amount and two currencies.
RiskMetrics allows for 30 different currency pairs (as measured against the USD).
Equity instruments are represented by an amount and currency of an equity basket
index in any of 30 different countries. Currently, RiskMetrics does not consider the
individual characteristics of a company stock but only the weighted basket of companies as represented by the local index.
Commodities positions are represented by amounts of selected standardized commodity futures contracts traded on commodity exchanges
3.

Software provided by J.P. Morgan, Reuters and third party firms that use the RiskMetrics
methodology and data documented herein.

Chart 1.9
Components of RiskMetrics

RiskMetrics
Volatility & correlation
estimates
Risk
Projection
Posting

Blotter
(Inventory)

Mapping

Estimated
Risks
Evaluation

Position
Valuation

Risk
/Return
Measures

Profits &
Losses

Transaction

RiskMetrics
methodology

System
implementations

Since the RiskMetrics methodology and the data sets are in the public domain and freely available,
anyone is free to implement systems utilizing these components of RiskMetrics. Third parties have
developed risk management systems for a wide range of clients using different methodologies.
The following paragraphs provide a taxonomy comparing the different approaches.

Part I: Risk Measurement Framework

18

Chapter 1. Introduction

1.3.2 Detailed specification

The section below provides a brief overview of how the RiskMetrics datasets are produced and
how the parameters we provide can be used in a VaR calculation.
1.3.2.1 Production of volatility and correlation data sets
RiskMetrics provides the following sets of volatility and corresponding correlation data files. One
set is for use in estimating VaR with a forecast horizon of one day. The other set is optimized for a
VaR forecast horizon of one month. The third set is based on the quantitative criteria set by the
Bank for International Settlements on the use of VaR models to estimate the capital required to
cover market risks. The process by which these data files are constructed are as follows:
1.

Financial prices are recorded from global data sources. (In 1997, RiskMetrics will switch to
using Reuters data exclusively). For certain fixed income instruments we construct zero
rates. See Chapter 9 for data sources and RiskMetrics building blocks.

2.

Fill in missing prices by using the Expectation Maximization algorithm (detailed in

Section 8.2). Prices can be missing for a variety of reasons, from technical failures to holiday schedules.

3.

Compute daily price returns on all 480 time series (Section 4.1).

4.

Compute standard deviations and correlations of financial price returns for a 1-day VaR
forecast horizon. This is done by constructing exponentially weighted forecasts. (See
Section 5.2). Production of the daily statistics also involves setting the sample daily mean
to zero. (See Section 5.3). If data is recorded at different times (Step 1), users may require
an adjustment algorithm applied to the correlation estimates. Such an algorithm is
explained in Section 8.5. Also, users who need to rebase the datasets to account for a base
currency other than the USD should see Section 8.4.

5.

Compute standard deviations and correlations of financial price returns for 1-month VaR
forecast horizon. This is done by constructing exponentially weighted forecasts
(Section 5.3). Production of the monthly statistics also involves setting the sample daily
mean to zero.

1.3.2.2 RiskMetrics VaR calculation

1. The first step in the VaR calculation is for the user to define three parameters:
(1) VaR forecast horizonthe time over which VaR is calculated, (2) confidence levelthe
probability that the realized change in portfolio will be less than the VaR prediction, and (3)
the base currency.
2.

For a given portfolio, once the cash flows have been identified and marked-to-market
(Section 6.1) they need to be mapped to the RiskMetrics vertices (Section 6.2).

3.

Having mapped all the positions, a decision must be made as to how to compute VaR. If the
user is willing to assume that the portfolio return is approximately conditionally normal,
then download the appropriate data files (instrument level VaRs and correlations) and compute VaR using the standard RiskMetrics approach (Section 6.3).

4.

If the users portfolio is subject to nonlinear risk to the extent that the assumption of conditional normality is no longer valid, then the user can choose between two methodologies
delta-gamma and structured Monte Carlo. The former is an approximation of the latter. See
Section 6.3 for a description of delta-gamma and Chapter 7for an explanation of structured
Monte Carlo.

RiskMetrics Technical Document

Fourth Edition

19

Chapter 2.

Historical perspective of VaR

2.1 From ALM to VaR
2.2 VaR in the framework of modern financial management
2.2.1
Valuation
2.2.2
Risk estimation
2.3 Alternative approaches to risk estimation
2.3.1
Estimating changes in value
2.3.2
Estimating market movements

Part I: Risk Measurement Framework

22
24
25
25
26
26
27

20

RiskMetrics Technical Document

Fourth Edition

21

Chapter 2.

Historical perspective of VaR

Jacques Longerstaey
Morgan Guaranty Trust Company
Risk Management Advisory
(1-212) 648-4936
[email protected]

Measuring the risks associated with being a participant in the financial markets has become the
focus of intense study by banks, corporations, investment managers and regulators. Certain risks
such as counterparty default have always figured at the top of most banks concerns. Others such
as market risk (the potential loss associated with market behavior) have only gotten into the limelight over the past few years. Why has the interest in market risk measurement and monitoring
arisen? The answer lies in the significant changes that the financial markets have undergone over
the last two decades.
1.

Securitization: Across markets, traded securities have replaced many illiquid instruments,
e.g., loans and mortgages have been securitized to permit disintermediation and trading.
Global securities markets have expanded and both exchange traded and over-the-counter
derivatives have become major components of the markets.
These developments, along with technological breakthroughs in data processing, have gone
hand in hand with changes in management practicesa movement away from management
based on accrual accounting toward risk management based on marking-to-market of positions. Increased liquidity and pricing availability along with a new focus on trading led to
the implementation of frequent revaluation of positions, the mark-to-market concept.
As investments became more liquid, the potential for frequent and accurate reporting of
investment gains and losses has led an increasing number of firms to manage daily earnings
from a mark-to-market perspective. The switch from accrual accounting to mark-to-market
often results in higher swings in reported returns, therefore increasing the need for managers to focus on the volatility of the underlying markets. The markets have not suddenly
become more volatile, but the focus on risks through mark-to-market has highlighted the
potential volatility of earnings.
Given the move to frequently revalue positions, managers have become more concerned
with estimating the potential effect of changes in market conditions on the value of their
positions.

2.

Performance: Significant efforts have been made to develop methods and systems to measure financial performance. Indices for foreign exchange, fixed income securities, commodities, and equities have become commonplace and are used extensively to monitor returns
within and/or across asset classes as well as to allocate funds.
The somewhat exclusive focus on returns, however, has led to incomplete performance
analysis. Return measurement gives no indication of the cost in terms of risk (volatility of
returns). Higher returns can only be obtained at the expense of higher risks. While this
trade-off is well known, the risk measurement component of the analysis has not received
broad attention.
Investors and trading managers are searching for common standards to measure market
risks and to estimate better the risk/return profile of individual assets or asset classes. Notwithstanding the external constraints from the regulatory agencies, the management of
financial firms have also been searching for ways to measure market risks, given the potentially damaging effect of miscalculated risks on company earnings. As a result, banks,
investment firms, and corporations are now in the process of integrating measures of market risk into their management philosophy. They are designing and implementing market
risk monitoring systems that can provide management with timely information on positions
and the estimated loss potential of each position.

Over the last few years, there have been significant developments in conceptualizing a common
framework for measuring market risk. The industry has produced a wide variety of indices to measure return, but little has been done to standardize the measure of risk. Over the last 15 years many
market participants, academics, and regulatory bodies have developed concepts for measuring

Part I: Risk Measurement Framework

22

Chapter 2. Historical perspective of VaR

market risks. Over the last five years, two approaches have evolved as a means to measure market
risk. The first approach, which we refer to as the statistical approach, involves forecasting a portfolios return distribution using probability and statistical models. The second approach is referred to
as scenario analysis. This methodology simply revalues a portfolio under different values of market rates and prices. Note that in stress scenario analysis does not necessarily require the use of a
probability or statistical model. Instead, the future rates and prices that are used in the revaluation
can be arbitrarily chosen. Risk managers should use both approachesthe statistical approach to
monitor risks continuously in all risk-taking units and the scenario approach on a case-by-case
basis to estimate risks in unique circumstances. This document explains, in detail, the statistical
approachRiskMetricsto measure market risk.
This chapter is organized as follows:
Section 2.1 reviews how VaR was developed to support the risk management needs of trading
activities as opposed to investment books. Though the distinction to date has been an accounting one not an economic one, VaR concepts are now being used across the board.
Section 2.2 looks at the basic steps of the risk monitoring process.
Section 2.3 reviews the alternative VaR models currently being used and how RiskMetrics
provides end-users with the basic building blocks to test different approaches.
2.1 From ALM to VaR
A well established method of looking at market risks in the banking industry is to forecast earnings under predetermined price/rate market conditions (or scenarios). Earnings here are defined as
earnings reported in a firms Financial Statements using generally accepted accounting principles.
For many institutions the bulk of activities are reported on an accrual basis, i.e., transactions are
booked at historical costs +/- accruals. Only a limited number of trading items are marked to market. Because changes in market rates manifest themselves only slowly when earnings are reported
on an accrual basis, the simulation of income has to be done over extended periods, i.e., until most
of the transactions on the books mature. Chart 2.1 illustrates this conventional Asset/Liability
Management approach.
Chart 2.1
Asset liability management
Inventory of
financial
transactions
Accrual
items

Income
simulation

Projected
income
statement

Trading
items
Intermediate term
rate forecasts

There are two major drawbacks to this methodology:

It requires projecting market rate developments over extended periods into the future.

RiskMetrics Technical Document

Fourth Edition

Sec. 2.1 From ALM to VaR

23

It supports the illusion that gains and losses occur at the time they show up in the accrual
accounts (i.e., when they are realized following accounting principles). What this means is
that return is only defined as net interest earnings, a framework which ignores the change in
price component of the return function.
Every investor would agree that the total return on a bond position is the sum of the interest
earned and the change in the value of the bond over a given time horizon. Traditional ALM, as
a result of accounting conventions, ignores the change in value of the instrument since positions are not marked to market. This has often lead crafty ALM managers to create positions
which look attractive on paper because of high net interest earnings, but which would not perform as well if their change in market value were considered.
The market risk in trading positions is usually measured differently and managed separately. Trading positions are marked-to-market and the market value is then subjected to projections of
changes in short term in rates and prices. This is much less hazardous as rate forecasts are usually
limited to short horizons, i.e., the time it should take to close out or hedge the trading position.
Chart 2.2
Value-at-Risk management in trading
Inventory of
financial
transactions
Accrual
items
Trading
items

mark to
market

Current market
rates & prices

Market
values

Value
simulation

Projected
market
value
changes

Short term
price forecasts

The distinction between accrual items and trading items and their separate treatment for market
risk management has led to significant complicationsparticularly when transactions classified as
trading items under generally accepted accounting principles are used to hedge transactions
classified as accrual items. In an effort to overcome this difficulty, many firms particularly
those with relatively large trading books have expanded the market risk approach to also include
accrual items, at least for internal risk management reporting. This is done by estimating the fair
market value of the accrual items and the changes in their fair value under different short term scenarios. Thus we are witnessing the evolution of an alternative to the conventional approach of
Asset/Liability Management, the Value-at-Risk approach. It started in pure trading operations, but
is now gaining increased following in the financial industry.

Part I: Risk Measurement Framework

24

Chapter 2. Historical perspective of VaR

Chart 2.3
Comparing ALM to VaR management
Conventional
Asset/Liability
Management

Projected
income
statement

Income
simulation

Intermediate term
rate forecasts

New
Value at Risk
Management

Inventory of
financial
transactions
Accrual
items
Trading
items

Proxy
values
mark to
market

Market
values

Current market
rates & prices

Risk factors

Projected
market
value
changes

Short term
price forecasts

The advantages of VaR Management are that it

Incorporates the mark-to-market approach uniformly.
Relies on a much shorter horizon forecast of market variables. This improves the risk estimate
as short horizon forecasts tend to be more accurate than long horizon forecasts.
Of course, drawbacks exist. One of them is that it may not be trivial to mark certain transactions to
market or even understand their behavior under certain rate environments. This is particularly true
for instruments such as demand deposits in a retail banking environment for example. Whatever
the difficulties, the aim of getting an integrated picture of a firms exposure to market risks is
worth a number of assumptions, some of which may be reasonable representations of reality. In the
case of demand deposits, a recent article by Professor Robert Jarrow outlines how power swaps
could be modelled to represent a retail banks core deposit base risks (RISK, February 1996).
Some critics also argue that marking-to-market all transactions over short time periods creates too
much earnings or volatility. Looking at risks in this fashion may be misleading. This is the direction of the industry and its accounting regulators however and it will be up to financial analysts to
adapt to the new environment. The volatility of earnings will not just appear out of the blue. The
changes in accounting practices will ultimately show economic reality as it really is.
Market risk can be absolute or relative. In its absolute form, what is measured is the loss in the
value of a position or a portfolio resulting from changes in market conditions. Absolute market
risk is what managers of trading operations measure. Corporates who wish to estimate real potential losses from their treasury operations also focus on absolute market risk. Regulators are interested in absolute market risks in relation to a firms capital. When investment performance is
measured against an index, the inherent market risk is relative in the sense that it measures the
potential underperformance against a benchmark.
2.2 VaR in the framework of modern financial management
As discussed before there are two steps to VaR measurement. First, all positions need to be marked
to market (valuation). Second we need to estimate the future variability of the market value.
Chart 2.4 illustrates this point.

RiskMetrics Technical Document

Fourth Edition

Sec. 2.2 VaR in the framework of modern financial management

25

Chart 2.4
Two steps beyond accounting

Current market
rates & prices

Projected scenarios or
estimated volatilities &
correlations

Valuation

Risk
Projection

Accounting
Accrual
items

Mapping

Equivalent
Position

Trading
items

Trading
items

Balance
Sheet

Economic
values

Total
Position
Mapping

Market
Risks

2.2.1 Valuation
Trading items are valued at their current prices/rates as quoted in liquid secondary markets. To
value transactions for which, in the absence of a liquid secondary market, no market value exists,
we first map them into equivalent positions, or decompose them into parts for which secondary
market prices exist. The most basic such part is a single cash flow with a given maturity and currency of the payor. Most transactions can be described as a combination of such cash flows and
thus can be valued approximately as the sum of market values of their component cash flows.
Only non-marketable items that contain options cannot be valued in this simple manner. For their
valuation we also need expected volatilities and correlations of the prices and rates that affect their
value, and we need an options pricing model. Volatilities describe potential movements in rates
with a given probability; correlations describe the interdependencies between different rates and
prices. Thus, for some valuations, we require volatilities and correlations.
2.2.2 Risk estimation
Here we estimate value changes as a consequence of expected changes in prices and rates. The
potential changes in prices are defined by either specific scenarios or a set of volatility and correlation estimates. If the value of a position depends on a single rate, then the potential change in value
is a function of the rates in the scenarios or volatility of that rate. If the value of a position depends
on multiple rates, then the potential change in its value is a function of the combination
of rates in each scenario or of each volatility and of each correlation between all pairs of rates.
Generating equivalent positions on an aggregate basis facilitates the simulation. As will be shown
later, the simulation can be done algebraically (using statistics and matrix algebra), or exhaustively
by computing estimated value changes for many combinations of rate changes.
In the RiskMetrics framework, forecasts of volatilities and correlations play a central role. They
are required for valuations in the case of derivatives, the critical inputs for risk estimation.

Part I: Risk Measurement Framework

26

Chapter 2. Historical perspective of VaR

2.3 Alternative approaches to risk estimation

More than one VaR model is currently being used and most practitioners have selected an
approach based on their specific needs, the types of positions they hold, their willingness to trade
off accuracy for speed (or vice versa), and other considerations.
The different models used today differ on basically two fronts:
How the changes in the values of financial instruments are estimated as a result of market
movements.
How the potential market movements are estimated.
What makes the variety of models currently employed is the fact that the choices made on the two
fronts mentioned above can be mixed and matched in different ways.
2.3.1 Estimating changes in value
There are basically two approaches to estimating how the value of a portfolio changes as a result
of market movements: analytical methods and simulation methods.
2.3.1.1 Analytical methods
One such method is the analytical sensitivity approach based on the following equation:
estimated value change = f (position sensitivity, estimated rate/price change)
where the position sensitivity factor establishes the relationship between the value of the instrument and of the underlying rate or price, and determines the accuracy of the risk approximation.
In its simplest form, the analytical sensitivity approach looks like this:
estimated value change = position sensitivity estimated rate change
For example, the value change of a fixed income instrument can be estimated by using the instruments duration. Although this linear approximation simplifies the convex price/yield relationship
of a bond, it is extensively used in practice because duration often accounts for the most significant percentage of the risk profile of a fixed income instrument. Similar simplifications can be
made for options where the estimated change in value is approximated by the options delta.
The initial versions of RiskMetrics basically used an analytical VaR approach that assumed that
market risk could be estimated by using a simple first-order approximation such as the one outlined above. We have since extended the analytical approach to account for nonlinear relationships
between market value and rate changes (e.g., options), which requires accounting for gamma risk
in addition to delta risk. The more refined version of the analytical approach looks like this:
estimated value change = (position sensitivity 1 estimated rate change)
+ 1/2 (position sensitivity 2) (estimated rate change)2 +...
In the case of an option, the first-order position sensitivity is the delta, while the second-order term
is the gamma. Higher order effects can also be estimated using an analytical approach, but the
math typically gets more complex.
The analytical approach requires that positions be summarized in some fashion so that the estimated rate changes can be applied. This process of aggregating positions is called mapping and is
described in Chapter 6.

RiskMetrics Technical Document

Fourth Edition

Sec. 2.3 Alternative approaches to risk estimation

27

The advantages of analytical models is that they are computationally efficient and enable users to
estimate risk in a timely fashion.
2.3.1.2 Simulation methods
The second set of approaches, typically referred to as Full Valuation models rely on revaluing a
portfolio of instruments under different scenarios. How these scenarios are generated varies across
models, from basic historical simulation to distributions of returns generated from a set of volatility and correlation estimates such as RiskMetrics. Some models include user-defined scenarios
that are based off of major market events and which are aimed at estimating risk in crisis conditions. This process is often referred to a stress testing.
Full Valuation models typically provide a richer set of risk measures since users are able to focus
on the entire distribution of returns instead of a single VaR number. Their main drawback is the
fact that the full valuation of large portfolios under a significant number of scenarios is computationally intensive and takes time. It may not be the preferred approach when the goal is to provide
senior management with a timely snapshot of risks across a large organization.
2.3.2 Estimating market movements
The second discriminant between VaR approaches is how market movements are estimated. There
is much more variety here and the following list is not an exhaustive list of current practice.
RiskMetrics
RiskMetrics uses historical time series analysis to derive estimates of volatilities and correlations
on a large set of financial instruments. It assumes that the distribution of past returns can be modelled to provide us with a reasonable forecast of future returns over different horizons.
While RiskMetrics assumes conditional normality of returns, we have refined the estimation process to incorporate the fact that most markets show kurtosis and leptokurtosis. We will be publishing factors to adjust for this effect once the RiskMetrics customizable data engine becomes
available on the Reuters Web.
These volatility and correlation estimates can be used as inputs to:
Analytical VaR models
Full valuation models. In Appendix E we outline how the RiskMetrics volatility and correlation data sets can be used to drive simulations of future returns.
Historical Simulation
The historical simulation approach, which is usually applied under a full valuation model, makes
no explicit assumptions about the distribution of asset returns. Under historical simulation, portfolios are valued under a number of different historical time windows which are user defined. These
lookback periods typically range from 6 months to 2 years.
Once the RiskMetrics customizable data engine becomes available on the ReutersWeb, users will
be able to access the underlying historical data needed to perform this type of simulation.
Monte Carlo Simulation
While historical simulation quantifies risk by replicating one specific historical path of market
evolution, stochastic simulation approaches attempt to generate many more paths of market
returns. These returns are generated using a defined stochastic process (for example, assume that
interest rates follow a random walk) and statistical parameters that drive the process (for example,
the mean and variance of the random variable).The RiskMetrics data sets can be used as inputs to
this process.

Part I: Risk Measurement Framework

28

Chapter 2. Historical perspective of VaR

In addition, the following VaR models add refinements to the results generated by the approaches
listed above.
Implied volatilities
Some practitioners will also look to the market to provide them with an indication of future potential return distributions. Implied volatility as extracted from a particular option pricing model is
the markets forecast of future volatility. Implied volatilities are often used in comparison to history to refine the risk analysis.
Implied volatilities are not currently used to drive global VaR models as this would require observable options prices on all instruments that compose a portfolio. Unfortunately, the universe of consistently observable options prices is not yet large enough; generally only exchange traded options
are reliable sources of prices. In particular, the number of implied correlations that can be derived
from traded options prices is insignificant compared to the number of correlations required to estimate risks in portfolios containing many asset types.
User-defined scenarios
Most risk management models add user-defined rate and price movements to the standard VaR
number, if only to test the effect of what could happen if historical patterns do not repeat themselves. Some scenarios are subjectively chosen while others recreate past crises events. The latter
is referred to as stress testing and is an integral part of a well designed risk management process.
Selecting the appropriate measurement method is not, however, straightforward. Judgment in the
choice of methodologies will always be important. Cost benefit trade-offs are different for each
user, depending on his position in the markets, the number and types of instruments traded, and the
technology available. Different choices can be made even at different levels of an organization,
depending on the objectives. While trading desks of a bank may require precise risk estimation
involving simulation on relatively small portfolios, senior management may opt for an analytical
approach that is cost efficient and timely. It is important for senior management to know whether
the risk of the institution is $10 million or $50 million. It is irrelevant for them to make the distinction between $10 million and $11 million. Achieving this level of accuracy at the senior management level is not only irrelevant, but can also be unachievable operationally, or at a cost which is
not consistent with shareholder value.
Since its introduction, RiskMetrics has become the umbrella name for a series of VaR methodologies, from the simple analytical estimation based on the precept that all instruments are linear (the
so-called delta approximation) to the structured Monte Carlo simulation.
Not all participants with exposure to the financial and commodities markets will have the
resources to perform extensive simulations. That is why we have strived in this update of the
RiskMetrics Technical Document to refine analytical approximations of risk for non-linear
instruments (the delta-gamma approximations). During 1997, the availability of historical rates
and prices under the RiskMetrics customizable data engine will make historical simulation an
option for users of our products.

RiskMetrics Technical Document

Fourth Edition

Sec. 2.3 Alternative approaches to risk estimation

29

Table 2.1
Two discriminating factors to review VaR models
How to estimate the change in the value of instruments
Analytical
How to estimate
rate and price
changes

Full VaR
model

Partial VaR
model

Full Valuation

RiskMetrics

Covariance matrices
applied to standard
instrument maps.

Covariance matrices used to define scenarios for

structured Monte Carlo.

Historical
simulation

Not applicable.

Portfolios revalued under historical return distributions (lookback period varies.

Monte Carlo

Not applicable.

Statistical parameters determine stochastic processes. Sources of data vary (can include
RiskMetrics covariance matrices).

Implied
volatilities

Covariance matrices
applied to standard
instrument maps.

Covariance matrices used to define scenarios for

structured Monte Carlo.

User defined

Sensitivity analysis
on single instruments.

Limited number of scenarios.

Part I: Risk Measurement Framework

30

Chapter 2. Historical perspective of VaR

RiskMetrics Technical Document

Fourth Edition

31

Chapter 3.

Applying the risk measures

3.1
3.2
3.3
3.4

Market risk limits

Calibrating valuation and risk models
Performance evaluation
Regulatory reporting, capital requirement
3.4.1
Capital Adequacy Directive
3.4.2
Basel Committee Proposal

Part I: Risk Measurement Framework

33
34
34
36
36
37

32

RiskMetrics Technical Document

Fourth Edition

33

Chapter 3.

Applying the risk measures

Jacques Longerstaey
Morgan Guaranty Trust Company
Risk Management Advisory
(1-212) 648-4936
[email protected]

The measures of market risk outlined in the preceding sections can have a variety of applications.
We will highlight just a few:
To measure and compare market risks.
To check the valuation/risk models.
To evaluate the performance of risk takers on a return/risk basis.
To estimate capital levels required to support risk taking.
3.1 Market risk limits
Position limits have traditionally been expressed in nominal terms, futures equivalents or other
denominators unrelated to the amount of risk effectively incurred. The manager of a USD bond
portfolio will be told for example that he cannot hold more than 100 million USD worth of U.S.
Treasury bonds. In most cases, the measure contains some risk constraint expressed in a particular
maturity or duration equivalent (if the 100 million limit is in 2-year equivalents, the manager will
not be able to invest 100 million in 30-year bonds). Setting limits in terms of Value-at-Risk has
significant advantages: position benchmarks become a function of risk and positions in different
markets while products can be compared through this common measure. A common denominator
rids the standard limits manuals of a multitude of measures which are different for every asset
class. Limits become meaningful for management as they represent a reasonable estimate of how
much could be lost.
A further advantage of Value-at-Risk limits comes from the fact that VaR measures incorporate
portfolio or risk diversification effects. This leads to hierarchical limit structures in which the risk
limit at higher levels can be lower than the sum of risk limits of units reporting to it.
Chart 3.1
Hierarchical VaR limit structure
Business Area
VaR-Limit:
$20MM

Business Group A
VaR-Limit:
$10MM

Unit A1
VaR-Limit:
$8MM

Business Group B
VaR-Limit:
$12MM

Unit A2
VaR-Limit:
$7MM

Business Group C
VaR-Limit:
$8MM

Unit C1
VaR-Limit:
$6MM

Unit C2
VaR-Limit:
$5MM

Unit C3
VaR-Limit:
$3MM

Setting limits in terms of risk helps business managers to allocate risk to those areas which they
feel offer the most potential, or in which their firms expertise is greatest. This motivates managers
of multiple risk activities to favor risk reducing diversification strategies.

Part I: Risk Measurement Framework

34

Chapter 3. Applying the risk measures

3.2 Calibrating valuation and risk models

An effective method to check the validity of the underlying valuation and risk models is to compare DEaR estimates with realized daily profits and losses over time. Chart 3.2 illustrates the concept. The stars show the daily P&L of a global trading business during the first 7 months of 1993,
the two lines show the Daily Earnings at Risk, plus and minus.
Chart 3.2
Ex post validation of risk models: DEaR vs. actual daily P&L
7.5

+ DEaR
5.0

-2.5

2.5

-5.0
- DEaR
-7.5
Jan

Feb

Mar

Apr

May

Jun

Jul

By definition, the cone delimited by the +/DEaR lines should contain 90% of all the stars,
because DEaR is defined as the maximum amount of expected profit or losses 90% of the time. If
there are substantially more than 10% of the stars outside the DEaR-cone, then the underlying
models underestimate the risks. If there are no stars outside the DEaR cone and not even close to
the lines, then the underlying models overestimate the risks.
This type of chart is only a reasonable reflection of the risk statistics if the daily profit and losses
are derived solely from overnight risk taking and not intraday trading and other activities. Often
this is not the case. Then instead of the daily P&L you should plot what is often referred to as the
no-action-P&L; it describes the hypothetical P&L on the position that would have been incurred
if the previous days closing position had been kept for the next 24 hours and then revalued. This
data is often difficult to collect.
3.3 Performance evaluation
To date, trading and position taking talent have been rewarded to a significant extent on the basis
of total returns. Given the high rewards bestowed on outstanding trading talent this may bias the
trading professionals towards taking excessive risks.It is often referred to as giving traders a free
option on the capital of your firm. The interest of the firm or capital provider may be getting out of
line with the interest of the risk taking individual unless the risks are properly measured and
returns are adjusted for the amount of risk effectively taken.

RiskMetrics Technical Document

Fourth Edition

Sec. 3.3 Performance evaluation

35

To do this correctly one needs a standard measure of risks. Ideally risk taking should be evaluated
on the basis of three interlinked measures: revenues, volatility of revenues, and risks. This is illustrated by Chart 3.3:
Chart 3.3
Performance evaluation triangle

Risks

Risk
Ratio

Efficiency
Ratio

Revenues

Volatility
of
revenues

Sharpe
Ratio

Including estimated (ex ante) and realized (ex post) volatility of profits adds an extra dimension to
performance evaluation. The ratio of P&L over risk (risk ratio) and of P&L over volatility (Sharpe
ratio) can be combined into what we define as a traders efficiency ratio (estimated risk/realized
volatility) that measures an individuals capacity to translate estimated risk into low realized volatility of revenues.
Consider an example to illustrate the issue. Assume you have to evaluate Trader #1 relative to
Trader #2 and the only information on hand is the history of their respective cumulative trading
revenues (i.e., trading profits). This information allows you to compare their profits and volatility
of their profits, but says nothing about their risks.
Chart 3.4
Example: comparison of cumulative trading revenues
cumulative revenues

Trader #2

Trader #1
6

1
0

-1

-1
1992

1993

Part I: Risk Measurement Framework

1992

1993

36

Chapter 3. Applying the risk measures

With risk information you can compare the traders more effectively. Chart 3.5 shows, for the two
traders the risk ratio, sharpe ratio, and efficiency ratio over time.
Chart 3.5
Example: applying the evaluation triangle

Trader #2

Trader #1
Sharpe
Ratio
P&L
vol(P&L)

Risk
Ratio
P&L
DEaR

0.6

0.6

-0.6

-0.6

0.4

0.4

0.2

0.2

0
-0.2

-0.2

Efficiency
Ratio
DEaR
vol(P&L)

0
1992

1993

1992

1993

Note, you have no information on the type of market these traders operate in or the size of positions they have taken. Nevertheless Chart 3.5 provides interesting comparative information which
lead to a richer evaluation.
3.4 Regulatory reporting, capital requirement
Financial institutions such as banks and investment firms will soon have to meet capital requirements to cover the market risks that they incur as a result of their normal operations. Currently the
driving forces developing international standards for market risk based capital requirements are
the European Community which issued a binding Capital Adequacy Directive (EC-CAD) and the
Basel Committee on Banking Supervision at the Bank for International Settlements (Basel Committee) which has recently come out with revised proposals on the use of banks internal models.
(See Appendix F for more information.)
3.4.1 Capital Adequacy Directive
The European Unions EEC 93/6 directive mandates banks and investment firms to set capital
aside to cover market risks. In a nutshell the EC-CAD computes the capital requirement as a sum
of capital requirements on positions of different types in different markets. It does not take into
account the risk reducing effect of diversification. As a result, the strict application of the current
recommendations will lead to financial institutions, particularly the ones which are active internationally in many different markets, to overestimate their market risks and consequently be required
to maintain very high capital levels. While there may be some prudential advantages to this, it is

RiskMetrics Technical Document

Fourth Edition

Sec. 3.4 Regulatory reporting, capital requirement

37

not an efficient allocation of financial resources and could lead certain activities to be moved outside the jurisdiction of the financial regulatory authorities.
3.4.2 Basel Committee Proposal
In January 1996, the Basel Committee on Banking Supervision of the BIS issued a revised consultative proposal on an Internal Model-Based Approach to Market Risk Capital Requirements that
represents a big step forward in recognizing the new quantitative risk estimation techniques used
by the banking industry. These proposals recognize that current practice among many financial
institutions has superseded the original guidelines in terms of sophistication, and that banks should
be given the flexibility to use more advanced methodologies. This so-called internal models
approach addresses a number of issues that were raised when banks commented on the original
proposal dated April 1993.
Table 3.1 compares the methodologies for estimating market risks as recently proposed by the
Basel Committee with the RiskMetrics methodology covered in this document. This comparison
focuses exclusively on the so-called quantitative factors that the BIS guidelines will require banks
to use. It does not address the qualitative ones related to the risk management process and which
are beyond the scope of this document.
While the methodologies outlined in the BIS proposals have come a long way in overcoming
important objections to the first set of proposals, there are still a number of issues that will be
debated further. In order to facilitate the discussion between regulators and regulated, we have
published since mid-1995 in parallel with the existing volatility and correlation data sets, a
RiskMetrics Regulatory Data Set. The distribution of this regulatory data set is not an endorsement
of the Basel committee proposals and the following paragraphs which explain how the data set can
be used do not constitute J.P. Morgans official position on the content and scope of the Basel committee proposal.
Consistent with the other RiskMetrics data sets, the Regulatory Data Set contains volatility estimates for a 1-day holding period. Given that the BIS rules require market risk estimates to be calculated over a 10-day holding period and a 99% confidence interval (i.e., 2.33 standard
deviations), users will need to rescale the 1-day volatility (see Eq. [3.1]). The Basel proposals
allow for this adjustment of data (they actually refer to scaling up VaR estimates but exclude this
practice in the case of options since it only works for instruments whose pricing formulae are linear). Scaling up volatility estimates is perfectly legitimate, assuming no autocorrelation in the
data. Scaling up Value-at-Risk does not work for options, though using scaled up volatilities to
estimate the market risks of options with adequate pricing algorithms poses no problem.
As in the other data sets, volatilities and correlations are measured as daily log changes in rates
and prices. However, contrary to the exponential weighting schemes used for the other data sets,
estimates in the Regulatory Data Set are based on simple moving averages of 1 year of historical
data, sampled daily.
To make it comparable to the standard data sets, the RiskMetrics Regulatory Data Set is based on
95% confidence. Including the adjustment for the holding period, users downloading the data sets
will need to rescale the volatility estimates according to the following equation, in order to meet
the requirements set forth in the Basel proposals (this adjustment assumes a normal distribution.
More refined methods incorporating the characteristics of fat tailed distributions are outlined in the
statistics section of this document):

[3.1]

2.33
10
V Basel = ---------- V
1.65 RiskMetrics RD
= 4.45 V RiskMetrics RD

Part I: Risk Measurement Framework

38

Chapter 3. Applying the risk measures

where
V RiskMetrics RD = volatilities provided in RiskMetrics Regulatory Dataset
V Basel = volatilities suggested by Basel Committee for use in internal models
Correlations across asset classes (i.e., foreign exchange to government bonds for example) are
supplied in the RiskMetrics Regulatory Data Set, despite the fact that actual use of empirical correlations in the VaR estimates is subject to regulatory approval. The BIS has stated that the use of
correlations across asset classes would be based on whether the supervisory authority was satisfied
with the integrity of the estimation methodology.

RiskMetrics Technical Document

Fourth Edition

Sec. 3.4 Regulatory reporting, capital requirement

39

Table 3.1
Comparing the Basel Committee proposal with RiskMetrics
Issue
Mapping:
how positions are described
in summary form

Volatility:
how statistics of future price
movement are estimated

Basel Committee proposal

RiskMetrics

Fixed Income: at least 6 time buckets, differentiate

government yield curves and spread curves.

Fixed Income: data for 710 buckets of government yield curves

in 16 markets, 4 buckets money market rates in 27 markets, 46
buckets in swap rates in 18 markets.

Equities: country indices, individual stocks on basis

of beta equivalent.

Equities: country indices in 27 markets, individual stocks on beta

(correction for non-systematic risk).

Commodities: to be included, not specified how.

Commodities: 80 volatility series in 11 commodities (spot and

term).

Volatility expressed in standard deviation of normal

distribution proxy for daily historical observations
year or more back. Equal weights or alternative
weighting scheme provided effective observation
period is at least one year.

Volatility expressed in standard deviation of normal distribution

proxy for exponentially weighted daily historical observations
with decay factors of .94 (for trading, 74 day cutoff 1%) and .97
(for investing, 151 day cutoff at 1%).
Special Regulatory Data Set, incorporating Basel Committee
1-year moving average assumption.

Adversity:
size of adverse move in terms
of normal distribution

Options:
treatment of time value and
non-linearity

Correlation:
how risks are aggregated

Residuals:
treatment of instrument
specific risks

Estimate updated at least quarterly.

Estimates updated daily.

Minimum adverse move expected to happen with

probability of 1% (2.32 standard deviations) over 10
business days. Permission to use daily statistics
scaled up with square root of 10 (3.1). Equivalent to
7.3 daily standard deviations.

For trading: minimum adverse move expected to happen with

probability of 5% (1.65 standard deviation) over 1 business day.

Risk estimate must consider effect of non-linear price

movement (gamma-effect).

Non-linear price movement can be estimated analytically (deltagamma) or under simulation approach. Simulation scenarios to
be generated from estimated volatilities and correlations.

Risk estimate must include effect of changes in

implied volatilities (vega-effect).

Estimates of volatilities of implied volatilities currently not provided, thus limited coverage of options risk.

Portfolio effect can be considered within asset classes

(Fixed Income, Equity, Commodity, FX). Use of correlations across asset classes subject to regulatory
approval.

Full portfolio effect considered across all possible parameter

combinations.

Correlations estimated with equally weighted daily

data for more than one year.

Correlations estimated using exponentially weighted daily historical observations with decay factors of 0.94 (for trading, 74 day
cutoff 1%) and 0.97 (for investing, 151 day cutoff at 1%).

Instrument specific risks not covered by standard

maps should be estimated.

Does not deal with specific risks not covered in standard maps.

Capital requirements at least equal to 50% of charge

calculated under standard methodology.

Part I: Risk Measurement Framework

For investment: minimum adverse move expected to happen with

probability of 5% (1.65 standard deviation) over 25 business
days.

40

Chapter 3. Applying the risk measures

RiskMetrics Technical Document

Fourth Edition