Ganapathy Vidyamurthy Pairs Trading

Commodity
For some, the trends - and volatility - in Subsequent chapters detailing risk management,
Commodity Investing and Trading

commodity markets in the 21st century trading, and market insights including:
can be summed up in one word: China. structural alpha strategies
energy index tracking
Investing
Yet those studying, trading and regulating enterprise risk management
these markets know that such deceptively CVA for commodity derivatives
simple descriptions cannot explain the subtle the future of markets in China.
dynamics that drive supply and demand.
Contributors include:
To be sure, Chinas growth, industrialisation Michael Haigh Socit Gnrale,
and Trading
and consumerism have led to soaring demand Kamal Naqvi Credit Suisse,
for everyday commodities: China now Mark Hooker State Street Global Advisors,
accounts for over 40% of the demand of the Carlos Blanco NQuantX, LLC and
worlds iron ore, copper, and other metals. Wang Xueqin Zhengzhou Commodity Exchange.
But commodity markets are now part of the Commodity markets are an indelible
integrated global financial system - buffeted element of financial markets and of society.
by demand from growing emerging-market For thousands of years they have shown
economies as much as by cash-rich funds themselves to be the most efficient way to
eyeing commodities as an asset class. assign the elemental resources
necessary to advance. This fundamental
Edited by Stinson Gibner

Editor Stinson Gibner brings two decades of quality has not changed.
experience to Commodity Investing and Trading, EDITED BY STINSON GIBNER
having cut his teeth at Enron, Citadel, and What has changed is the breadth,
Citigroup. He has assembled a team of industry depth and complexity of markets.
experts whose contributions give the reader
a unique view of the commodity markets.
Chapters focus on the fundamentals

of major, key markets:
oil and petroleum
metals
natural gas
power
weather
grains and oilseeds
coal.
PEFC Certified
This book has been

produced entirely from
sustainable papers that
are accredited as PEFC
compliant.
www.pefc.org
Commodity Investing and Trading.indd 1 01/10/2013 15:36

00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page i

00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page ii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page iii
Stinson Gibner
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page iv
Published by Risk Books, a Division of Incisive Media Investments Ltd
Incisive Media
3234 Broadwick Street
London W1A 2HG
Tel: +44(0) 20 7316 9000
E-mail: [email protected]
Sites: www.riskbooks.com
www.incisivemedia.com
2013 Incisive Media
ISBN 978 1 906348 84 7
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
Publisher: Nick Carver

Associate Editor: Alice Levick
Managing Editor: Lewis OSullivan
Designer: Lisa Ling

Copy-edited by Laurie Donaldson
Typeset by Mark Heslington Ltd, Scarborough, North Yorkshire

Printed and bound in the UK by Berforts Group Ltd
Conditions of sale
All rights reserved. No part of this publication may be reproduced in any material form whether
by photocopying or storing in any medium by electronic means whether or not transiently or
incidentally to some other use for this publication without the prior written consent of the
copyright owner except in accordance with the provisions of the Copyright, Designs and Patents
Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Limited of
Saffron House, 610 Kirby Street, London EC1N 8TS, UK.
Warning: the doing of any unauthorised act in relation to this work may result in both civil and
criminal liability.
Every effort has been made to ensure the accuracy of the text at the time of publication, this
includes efforts to contact each author to ensure the accuracy of their details at publication is
correct. However, no responsibility for loss occasioned to any person acting or refraining from
acting as a result of the material contained in this publication will be accepted by the copyright
owner, the editor, the authors or Incisive Media.
Many of the product names contained in this publication are registered trade marks, and Risk
Books has made every effort to print them with the capitalisation and punctuation used by the
trademark owner. For reasons of textual clarity, it is not our house style to use symbols such as
TM, , etc. However, the absence of such symbols should not be taken to indicate absence of
trademark protection; anyone wishing to use product names in the public domain should rst
clear such use with the product owner.
While best efforts have been intended for the preparation of this book, neither the publisher, the
editor nor any of the potentially implicitly afliated organisations accept responsibility for any
errors, mistakes and or omissions it may provide or for any losses howsoever arising from or in
reliance upon its information, meanings and interpretations by any parties.
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page v
Contents
About the Editors ix

About the Authors xi
Introduction xvii
PART I: COMMODITY MARKET FUNDAMENTALS
1 The Impact of Non-fundamental Information on Commodity

Markets 3
Michael S. Haigh
Socit Gnrale Corporate and Investment Bank
2 The North American Natural Gas Market 25

Stinson Gibner
Whiteside Energy
3 A Day in the Life of Commodity Weather 65

Jose Marquez
Whiteside Energy
4 Oil and Petroleum Products: History and Fundamentals 75

Todd J. Gross
QERI LLC
5 Wholesale Power Markets 113

William Webster
RWE Supply and Trading
6 The Metals Markets 133

Kamal Naqvi
Credit Suisse
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page vi
COMMODITY INVESTING AND TRADING
7 Grains and Oilseeds 165

David Stack
Agrimax
8 Coal 207
Jay Gottlieb
PART II: TRADING AND INVESTMENT STRATEGIES
9 Farmland as an Investment 229

Greyson S. Colvin and T. Marc Schober
Colvin & Co. LLP
10 Agriculture Trading 249

Patrick OHern
Sugar Creek Investment Management
11 Quantitative Approaches to Capturing Commodity Risk

Premiums 295
Mark Hooker and Paul Lucek
State Street Global Advisors and SSARIS Advisors
12 Structural Alpha Strategies 307

Francisco Blanch; Gustavo Soares and Paul D. Kaplan
Bank of America Merrill Lynch; Macquarie Funding
Holding Inc. and Morningstar, Inc.
13 Energy Index Tracking 337

Kostas Andriosopoulos
ESCP Europe Business School
PART III: MARKET DEVELOPMENTS AND RISK MANAGEMENT
14 Enterprise Risk Management for Energy and Commodity

Physical and Financial Portfolios 371
Carlos Blanco
NQuantX LLC and MTG Capital Management
vi
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page vii
CONTENTS
15 Credit Valuation Adjustment (CVA) for Energy and

Commodity Derivatives 389
Carlos Blanco; and Michael Pierce
NQuantX LLC and MTG Capital Management; NQuantX LLC
16 The Past, Present and Future of Chinas Futures Market:

Trading Volume Analysis 409
Wang Xueqin
Zhengzhou Commodity Exchange
Index 439
vii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page viii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page ix
About the Editor
Stinson Gibner is an analyst at Whiteside Energy, having worked in

energy risk management and trading since the early 1990s. He previ-
ously headed the quantitative analytics team as a managing director
for Citigroup Global Commodities, supporting offices in Houston,
London and Singapore. Before joining Citigroup in 2005, Stinson
served as a director at Citadel Investment Group LLC, where he was
responsible for developing models and systems used for energy
trading and risk management. Between 1992 and 2001, he worked in
the quantitative modelling group at Enron Corp. Stinson received his
BA in physics from Rice University and a PhD from Caltech.
ix
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page x
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xi
About the Authors
Kostas Andriosopoulos is executive director of the Research Centre

for Energy Management at ESCP Europe Business School. His
research interests include price modelling, financial engineering and
the application of risk management techniques and innovative
investment strategies in energy, shipping and agricultural commodi-
ties markets, and international trade. Kostas is the associate editor for
the International Journal of Financial Engineering and Risk Management
and has organised numerous international conferences. He holds a
PhD in finance from Cass Business School, London, an MBA and
MSc in finance from Northeastern University, Boston, and a bach-
elors degree in production engineering and management from the
Technical University of Crete, Greece.
Francisco G. Blanch is managing director and head of global

commodities and derivatives research at Bank of America Merrill
Lynch, where he is also a member of the research investment and
executive management committees. Prior to joining Merrill Lynch,
he was an energy economist at Goldman Sachs and consulted for the
European Commission. Francisco holds a doctorate in economics
from Complutense University of Madrid and a masters in public
administration from Harvard University, where he was also a
teaching fellow in financial markets.
Carlos Blanco is managing director of NQuantX LLC, and director of

risk management at MTG Capital. He is also a faculty member at The
Oxford Princeton Programme, where he heads the Certificate
Programme on Derivatives Pricing, Hedging and Risk Management.
Greyson S. Colvin is founder and managing partner of Colvin & Co,

an agriculture-focused investment manager. Previously, he was a
research analyst at Credit Suisse in the Portfolio Management Group
and at UBS Investment Research. Greyson has been featured in
xi
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xii
numerous publications and is co-author of the Investors Guide to

Farmland. He received a BA in financial management from the
University of St. Thomas and an MBA in finance and investment
banking from the University of Wisconsin, Madison.
Rita DEcclesia is a professor at Sapienza University of Rome and

visiting professor at Birkbeck College, University of London. She is
also a director of the PhD programme in Economics and Finance at
Sapienza, as well as the director of the International Summer School
on Risk Measurement and Control, chair of the Euro Working Group
for Commodities and Financial Modeling and associate editor of
several scientific journals. Rita teaches courses at graduate and PhD
levels on quantitative models, finance and asset pricing. Rita's
research activity focuses on optimisation techniques and modelling
financial and energy commodity markets. She is active within the
Research Centre for Energy Management at ESCP Europe.
Jay Gottlieb led development of the first coal derivatives instru-

ment, the NYMEX CAPP coal futures contract, while a director in the
Exchange's Research Department. Jay was also instrumental in the
launch of instruments and over the counter clearing for the electricity
and emissions markets, and exchange traded funds for gold and oil
markets. He has served as a member of the board of directors of the
New York State Energy Research and Development Authority and
the Coal Trade Association. He holds an MBA from Stanford, a BS
from Huxley College of the Environment, and a BA from St. John's
College, Annapolis.
Todd Gross is chief investment officer, managing member and

founder of QERI LLC, a New York commodity trading firm which
invests client assets in liquid, fundamentally-based strategies.
Throughout a 25-year career Todd has been dedicated to under-
standing the nuances and inefficiencies of the commodity space with
particular emphasis in Energy. He began his career at Cooper Neff &
Associates, moved on to manage derivatives in Morgan Stanley's
Global Commodity Group, and founded and ran Hudson Capital
Group LLC, before launching QERI LLC in 2012. Todd received a BS
in economics from Wharton and a bachelor of applied science in
systems engineering from the Moore School of Engineering.
xii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xiii
ABOUT THE AUTHORS
Michael Haigh is managing director and global head of commodi-

ties research for Socit Gnrale, based in New York City, managing
a team of commodity analysts in Singapore, Paris, London and New
York City. Prior to joining Socit Gnrale, he was global head of
commodities research at Standard Chartered Bank in Singapore.
Michael has also held the position of managing director at K2
Advisors, and spent several years as the associate chief economist at
the US Commodity Futures Trading Commission and as a tenured
associate professor of economics at the University of Maryland. He
holds a PhD in economics with a minor in statistics from North
Carolina State University.
Mark Hooker was most recently senior managing director of State

Street Global Advisors and head of its Advanced Research Center,
where he was responsible for the worldwide development and
enhancement of SSgAs quantitative investment models. Prior to
joining SSgA in 2000, Mark was a financial economist with the
Federal Reserve Board in Washington, and before that an assistant
professor of economics at Dartmouth College. He earned a PhD in
economics from Stanford University and a bachelors degree with a
dual concentration in economics and mathematics from the
University of California at Santa Barbara.
Paul D. Kaplan is director of research for Morningstar Canada and a

senior member of Morningstars global research team, as well as a
qualified CFA. He is responsible for many of the quantitative
methodologies behind Morningstars fund analysis, indexes, advisor
tools and other services. Pauls research has appeared in many
professional publications, including his book, Frontiers of Modern
Asset Allocation. He received his bachelors degree from New York
University and his masters and doctorate in economics from
Northwestern University.
Paul R. Lucek is the chief investment officer, Hedge Fund Group,

and a member of the Hedge Fund Investment Committee at SSARIS
Advisors. Prior to joining SSARIS, he developed quantitative algo-
rithms for trading stock index futures, and in 1996 he co-founded
SITE Capital Management. He made the transition to money
management from the MD/PhD programme at Columbia
xiii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xiv
University, College of Physicians and Surgeons, where as a

researcher he pioneered the use of neural networks in the analysis of
complex genetic inheritance in humans. Paul earned his bachelors
and masters degrees in biology from Harvard University, and a
masters degree in genetics from Columbia University.
Jose Marquez is a meteorologist for Whiteside Energy. Since 2000,

his meteorology experience has been focused on the energy industry,
where he has held positions as senior meteorologist at Total Gas &
Power, Citigroup, Citadel Investment Group and Enron North
America. After graduating from the Navys Meteorological and
Oceanographic training school, he served in the US Navy, and he
was also director of meteorology for the Latin America Weather
Channel. He has a BS in environmental sciences from the University
of Puerto Rico and an MS in atmospheric sciences from the Georgia
Institute of Technology.
Kamal Naqvi is a managing director, global head of metals and head

of commodity sales across Europe, the Middle East and Africa in the
investment banking division of Credit Suisse, based in London. He
has been working in the resources industry since the early 1990s,
having also worked in commodity sales and commodity research
positions at Barclays Capital, Macquarie Bank and the Tasmanian
State Government. Kamal holds degrees in law and in economics
(hons) from the University of Tasmania.
Patrick E. O'Hern is the managing partner and co-founder of Sugar

Creek Investment Management, an actively managed commodity
trading and alternative investments advisor in Chicago. Patrick is
also head of portfolio management for the Meech Lake Investment
Group, a commodity trading asset manager. Previously, Patrick
held the position of senior analyst in the funds group at FourWinds
Capital Management in Boston. Prior to joining FourWinds Patrick
spent his early career in trading and brokerage on the floor of the
Chicago Mercantile Exchange, where he traded in the livestock and
dairy pits. Patrick has a bachelor's degree in agriculture business
from Western Illinois University.
xiv
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xv
INTRODUCTION
Michael Pierce is co-founder and director of Financial Engineering

at NQuantX LLC a financial engineering firm which develops soft-
ware for portfolio valuation and risk management. He also worked
with Platts as the lead financial engineer and analytics software
developer. He is a former senior financial engineer at Financial
Engineering Associates (a MSCI/Barra company), where he was
responsible for front-line development of numerous software prod-
ucts over an eight-year period. Michael has a master's degree in
mathematics from the University of California at Berkeley.
T. Marc Schober is a director at Colvin & Co and managing editor of

Farmland Forecast. He has been featured in numerous publica-
tions and is co-author of the Investors Guide to Farmland. Growing up
on a Wisconsin farm, the Schober family has owned and managed
farmland in Wisconsin for over 40 years. He received a BS in business
management from the University of Wisconsin, Eau Claire, and is
also involved in a number of cancer fundraisers, including the
Oconomowoc LakeWalk.
Gustavo Soares is part of the Commodity Investor Products Group

at Macquarie Bank, where he is responsible for designing investable
strategies and indexes in commodities. He joined Macquarie in 2012,
having spent several years at Bank of America Merrill Lynch
working as a commodity strategist. Gustavo holds a BA/MA in
economics from Universidade de So Paulo, Brazil and a PhD in
economics from Yale University.
David G. Stack is managing director of Agrimax, a commodity

market consulting firm. He has worked in the commodities industry
since the late 1980s on all aspects of the energy and agricultural
markets. David is experienced in all parts of the physical and finan-
cial space, and specialises in derivatives, with clients ranging from
the smallest producer to the largest consumer, including hedge
funds and NGOs. Having previously worked at Barclays, Louis
Dreyfus, Bunge, Enron and BP, he is also MD of the Commodity
Trading Room at ESCP Europe, and develops trading and risk
management software with riskGRID.
xv
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xvi
William Webster is head of EU power market design for RWE

Supply and Trading. He began his career in the UK Government
Economic Service, ending with a period at UK water regulator
Ofwat, where he was a team leader. William joined the European
Commission in 2000, working in both DG Energy and Competition,
and introduced competition into electricity and gas markets. In 2007,
he joined RWE and ran two major strategy projects for RWE power
before starting his current role in 2010. William read economics at
Cambridge University, has an MA from the College of Europe and is
a member of the Chartered Institute for Securities and Investment.
Wang Xueqin is a senior specialist of the Zhengzhou Commodity

Exchange, where his major research areas are market development,
new products and commodity options. He previously worked for
the International Department of China Securities Regulatory
Commission, as well as the working taskforce for Chinas prepara-
tions for launching CIS 300 at CFFEX. Wang was the first from
Chinas futures industry to research options as a visiting scholar at
CBOE and IIT, and he has worked for Zhengzhou Grain Wholesale
Market, the precursor of the China Zhengzhou Commodity
Exchange.
xvi
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xvii
Introduction
Stinson Gibner
Whiteside Energy
Strong gains in commodity prices since the early 2000s created a

growing interest in the asset class. The financial industry responded
with many products, including new hedge funds, index funds,
commodity-linked fixed income products and exchange-traded
funds (ETFs). With oil and natural gas making a prominent peak in
2008 and gold hitting a peak in 2011, many took this as a sign that the
commodity bull had run its course and expected that we would
return to the normal long-standing trend of commodity price defla-
tion.
The deflationist camp notes that growth in China must slow
down, possibly to a dramatic degree, if imbalances in that economy
are not handled carefully. Europe and the US continue their struggle
to reignite sorely needed jobs growth in order to relieve high youth
unemployment, while at the same time facing demographics that
lead to a shrinking labour force.
However the worlds economic situation is resolved, commodities
and commodity flows will remain critical to the functioning of cities,
states and economies. For this reason, a basic knowledge of the
supply and demand issues relevant to each commodity sector
provides financial insights even beyond the commodity markets.
This book therefore discusses the fundamentals of many of the major
traded commodities offering both an introduction and a reference
for all those interested in understanding and analysing these
markets.
This book is divided into three sections. The first covers the funda-
mentals of the most important markets in energy, metals and
xvii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xviii
agriculture. Michael S. Haigh starts us off with an investigation into

the importance of non-fundamental information. He uses principle
component analysis to discover how commodity market behaviour
has changed over the years, and shows evidence that commodity
market participants have adjusted their behaviour since the financial
crisis of 200708.
Within the energy complex, crude oil, European power, North
American power, natural gas, liquefied natural gas (LNG), and coal
are covered in separate chapters. Chapter 2 by Stinson Gibner
provides an introduction to the fundamentals of the North American
market for natural gas. Natural and economic forces impacting
supply are illustrated along with the annual rollercoaster of demand.
The critical role of storage in balancing short-term and seasonal
swings is explained, and key issues for the supplydemand balance
are discussed. Also within this chapter, Rita D'Ecclasia gives an
overview of the expanding global LNG trade.
Relevant to all commodities, Jose Marquez discusses weather and
climate from the unique perspective of a working commodities mete-
orologist. His chapter walks through the daily analysis and
information flow that must be monitored to keep abreast of weather
impacts on commodity demand and supply, while a panel discusses
climatology and its longer-term indicators of weather trends.
In Chapter 4, Todd Gross tells the incredible story of how oil prices
climbed from US$17/bbl in 2002 to an amazing US$147/bbl a mere
six years later. Todd also examines, as he puts it, why the globe
always seems to be running out of oil, and yet, so far, that fate has yet
to be realised. Unafraid of digging into the details, the analysis of
global refining capacity gives a great insight into the changing
demand for and flows of various types of crude. The impact of
transport bottlenecks within North America is also discussed.
William Webster then explains the unique challenges of operating
a market for power, and explores the solution adopted by the
European power market. He explains the instruments traded and
price formation, before offering a historical perspective of pricing for
these markets. He concludes with thoughts about possible future
market trends and regulatory changes.
Kamal Naqvi takes us on a whirlwind tour through the precious,
base and industrial metals in Chapter 6, in which he discusses the
key drivers for metals and offers insights as to which may outper-
xviii
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xix
INTRODUCTION
form going forward. In Chapter 7, David Stack provides a tour de

force discussion of the global grains markets, giving an overview of
the markets for food grains, feed grains and oilseeds. Farming area,
yield and production trends are discussed for major producers. The
chapter also reviews past import and export flows, as well as likely
future trends for global grain flows and crop rotation flexibility. A
discussion of the coal markets in Chapter 8 completes the energy
commodities. Jay Gottlieb lays out the fundamentals of the coal
markets and discusses which trends are likely to dominate going
forward.
Rounding out agricultural investments, Greyson Colvin and Marc
Schober open the second section of the book by explaining the basics
of agricultural land in Chapter 9, arguing that the fundamentals
behind the rush to invest in farmland are likely to persist far into the
future. Complementing the grains discussion, Chapter 10 by Patrick
OHern explores the agricultural trading and hedging markets and
gives an overview of the types of participants active. He provides
several examples of trading strategies to illustrate intra-market and
cross-market trade opportunities within the agriculture markets,
and illustrates the diversification that may be provided across
commodities.
The remainder of the section on trading and investing strategies
focuses on alpha strategies and index investing. In Chapter 11, Mark
Hooker and Paul Lucek present an interesting case study of what
they call convergent and divergent strategies, concluding that useful
risk diversification can be achieved through intelligent choice of
strategies within a commodity portfolio. An overview of alpha
strategies that could be used by either traders or index funds is given
by Francisco Blanch and Gustavo Soares in Chapter 12, which covers
momentum, roll yield and volatility methods. The accompanying
panel by Paul Kaplan gives a short case study of active index funds
applying these alpha concepts.
In Chapter 13, Kostas Andriosopoulos finishes out the commodity
index investing discussion by bridging the gap between commodi-
ties and equities. It presents his proposal to track a spot commodity
index by using a carefully selected portfolio of equities, and shows
his tested selection methods and tracking results.
The third section of the book opens with two chapters by Carlos
Blanco. In Chapter 15, Blanco and Michael Pierce present the choices
xix
00 Prelims CIT_Commodity Investing and Trading 26/09/2013 11:30 Page xx
for performing a proper analysis of credit risks embedded in your

bilateral trade portfolio. The resulting credit value adjustment (CVA)
provides a measure of expected future loss due to credit events.
Taking a broader view of risk, Chapter 14, also written by Carlos
Blanco, explains the structure for putting in place a system of enter-
prise risk management and some possible pitfalls. In principle,
everyone wants to have proper risk systems and structures in place;
however, operational weakness is difficult to avoid as daily choices
must be made between risk levels and the potential profitability of
the enterprise. Carlos explains the challenges of the risk manager
and offers advice about properly structuring a risk management
function.
Wang Xueqin then reviews the rapid growth of commodities
trading in China in Chapter 16, and shows that although still largely
restricted to domestic participants, the size of Chinas commodity
futures markets now rivals commodity markets globally.
xx
01 Chapter CIT_Commodity Investing and Trading 25/09/2013 15:43 Page 1
Part I
Commodity Market
Fundamentals
1
The Impact of Non-fundamental
Information on Commodity Markets
Michael S. Haigh
Socit Gnrale Corporate and Investment Bank
Commodity markets can (and will) occasionally co-move with

broader macro markets for reasons beyond the physical fundamen-
tals. The purpose of this chapter is to illustrate how different
commodities are affected by non-fundamental factors (macro shocks,
liquidity events, currency moves and broader market sentiment
swings) that are normally considered exogenous to commodity
fundamentals (eg, mine or oil supply). At certain points in time, espe-
cially since the Lehman bankruptcy in September 2008, the
non-fundamental influences on certain commodities have dwarfed
the impact of actual commodity fundamentals. Accordingly, under-
standing this has brought obvious benefits for analysing how
commodity market price moves can be applied to trading strategies.
The chapter will examine this by focusing on energy (oil), base metals
(copper) and precious metals (gold), and agriculture (soybeans).
Until the late 1990s, commodity markets generally enjoyed excess
capacity as innovations and new discoveries resulted in greater
supply (think of how the US natural gas markets have evolved). Any
supply side shock that was persistent would result in commodity
price increases, higher inflation and a consequent decline in equity
markets: hence the negative relationship with commodity price
movements. However, by 2000, increased demand growth and
underinvestment in the supply chain meant the excess capacity was
slowly absorbed. By 2008, the underinvestment in activity became
more evident as the credit crisis resulted in suspensions and cancella-
3
tions of hundreds of commodity projects. As such, changes in global

economic activity, Purchasing Managers Index (PMI) strength, dollar
strength and changes in inflation expectations (resulting from quanti-
tative easing) are playing a much larger role in commodities.
Given the increased influence of non-fundamental information
on commodity markets, it is worthwhile to quantify as accurately as
possible the level of this influence, across commodities and across
time. To thoroughly assess the role of non-fundamentals during
episodes of quantitative easing, we employ the principal component
analysis (PCA) technique. Simply stated, PCA is the analysis of the
covariance matrix and can be used to analyse multi-assets: baskets
made up of commodities, other financial indicators, volatilities, etc.
From the historical data, the analysis determines the principal
components of the covariance matrix ie, the way in which the asset
price movements correlate, by order of importance. To conduct PCA
on commodity prices, we analyse price data for a variety of
commodities against a diversified basket of 28 assets across markets,
including: volatility indexes (EU and US), credit (EU and US and HY
versus IG), FX (dollar, yen, euro, carry trade (G10 and EM)), bonds
(spreads, 10Y GVT and inflation break-even), equities (BRIC, Euro,
emerging, EU and US) and global indexes. Factors are estimated
using the 28-member basket, which means each factor is a weighted
average of the 28 assets with different weights for each factor.
The model estimated three main explanatory factors: macro,
dollar and liquidity. What is not explained by these factors (the resid-
uals) is interpreted as the commodity fundamentals. Depending on
the commodity, the relative importance of each factor varies consid-
erably, as does the influence on commodities of all the factors
combined. Moreover, extreme events (eg, Lehman Brothers bank-
ruptcy) have structurally altered the influence of the macro factor (in
particular) and largely demoted the dollar factor to a secondary
outside influence on the commodity markets.
Energy
Here we focus on Brent and note that, unsurprisingly, Brents funda-
mentals in terms of explanatory power began to deteriorate
(consistently) in 2007 when the subprime crisis became a reality (see
Figure 1.1). In the early 2000s, fundamentals prevailed with 8090%
explanatory power (eg, in March 2002, dollar and liquidity explained
4
THE IMPACT OF NON-FUNDAMENTAL INFORMATION ON COMMODITY MARKETS
roughly 10% each of Brents price movements). The Lehman bank-

ruptcy changed this, with fundamentals explanatory power
dropping to the 3040% range, on average. Since 2013, we have seen
Brents fundamentals progressively giving up explanatory power to
the macro influences as inventories increase, alleviating concerns of a
shortage (see Figure 1.2). The dollars influence has latterly been
practically irrelevant in determining the price path for Brent.
Figure 1.1 Non-fundamentals influence on Brent experienced a structural break

in 2008, jumping from 2030% to over 80%
Macro Dollar Liquidity
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Jul-08
Jul-09
Jul-10
Jul-11
Jul-12
Jan-08
Jan-09
Jan-10
Jan-11
Jan-12
Jan-13
Source: SG Cross Asset Research
Figure 1.2 In late 2012 and early 2013 the macro influences had taken away
from Brents fundamentals
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Mar-11
Mar-02
Mar-03
Mar-04
Mar-05
Mar-06
Mar-07
Mar-08
Mar-09
Mar-10
Mar-12
Mar-13
Sep-11
Sep-02
Sep-03
Sep-04
Sep-05
Sep-06
Sep-07
Sep-08
Sep-09
Sep-10
Sep-12
5
Base
Prior to the Lehman crisis, the bulk of the explanatory power relating
to base metals price movement was explained by fundamentals (for
both copper and aluminium (not shown)), followed by movements
in the dollar. The role of fundamentals diminished post-Lehman
with more explanatory power coming from the macro factors and
much less from the dollar (see Figure 1.3). Copper is the one base
metal that is very exposed to the macro outlook, especially as price
levels have become significantly higher than the marginal cost of
production. Not surprisingly, prices can be significantly influenced
by other factors. In late 2012, the role of macro dropped in its
explanatory power (Figure 1.4).
Precious
The gold market remains an outlier among commodities (not
surprisingly), with the influence from non-fundamentals still coming
from the dollar, and liquidity and macro factors jostling for second
place in terms of explanatory. Since Lehman (Figure 1.5), liquidity
has improved in terms of extra explanatory power of gold price
movements. Since late 2012, the outside influences have dimin-
ished (see Figure 1.6), coinciding with gold prices plummeting in
early April 2013.
Figure 1.3 Copper the dollar has taken a back seat to macro since Lehman
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Mar-11
Mar-02
Mar-03
Mar-04
Mar-05
Mar-06
Mar-07
Mar-08
Mar-09
Mar-10
Mar-12
Mar-13
Sep-11
Sep-02
Sep-03
Sep-04
Sep-05
Sep-06
Sep-07
Sep-08
Sep-09
Sep-10
Sep-12
6
Figure 1.4 The role of macro has deteriorated since late 2012
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Jul-11
Jul-08
Jul-09
Jul-10
Jul-12
Jan-08
Jan-09
Jan-10
Jan-11
Jan-12
Jan-13
Figure 1.5 Gold: the dollar is usually the greatest influence
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Mar-11
Mar-02
Mar-03
Mar-04
Mar-05
Mar-06
Mar-07
Mar-08
Mar-09
Mar-10
Mar-12
Mar-13
Sep-11
Sep-02
Sep-03
Sep-04
Sep-05
Sep-06
Sep-07
Sep-08
Sep-09
Sep-10
Sep-12
Agriculture
The markets fundamentals (here represented by soybeans)
accounted for approximately 7095% of price volatility prior to
Lehman (see Figure 1.7). The remainder of the price movement was
captured mainly by the dollar (after the early 2000 recession).
Nevertheless, soybeans could not avoid the influence of the Lehman
7
Figure 1.6 Outside influences on gold have become irrelevant since late 2012
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Jan -08
Jan -09
Jan -11
Jan -12
Jan -13
Jan -10
Figure 1.7 Percentage explanation: fundamentals versus non-fundamentals;

soybean fundamentals have been resilient over the years
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Mar-11
Mar-02
Mar-03
Mar-04
Mar-05
Mar-06
Mar-07
Mar-08
Mar-09
Mar-10
Mar-12
Mar-13
Sep-11
Sep-02
Sep-03
Sep-04
Sep-05
Sep-06
Sep-07
Sep-08
Sep-09
Sep-10
Sep-12
crisis, as the percentage explanation coming from the macro factors

increased immediately following that event. Since late 2012, soybean
fundamentals have returned, explaining almost 100% of the price
move (Figure 1.8).
In summary, the more supply constraints, the lower the invento-
ries, the closer the price to the marginal cost of production and the
8
Figure 1.8 The drought of 2012 brings more explanatory power from soybean
fundamentals on price movement
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
1
8
2
11
8
3
l-1
l-0
l-0
l-1
-1
-0
-0
-1
-1
-1
n-
l
n
n
Ju
Ju
Ju
Ju
Ju
Ja
Ja
Ja
Ja
Ja
Ja
lesser the impact of outside factors on commodity markets.

Agriculture continues to be the most independent of the markets
(alongside natural gas), relying mainly on its own fundamentals.
Structurally, we have seen a shift in all markets (except gold)
whereby the role of the US dollar in terms of explanatory power has
dropped dramatically, to be replaced by the role of macro factors.
THE SG SENTIMENT INDICATOR VERSUS COMMODITIES

The job of assessing commodity price movements becomes difficult
when macro, dollar and liquidity dominate. It becomes even more
difficult when prices are pulled around by market sentiment.
Fortunately, we can assess the role of sentiment employing a senti-
ment indicator a tool used to gauge an average level of risk
experienced throughout the global markets. Although the method-
ology is intuitive and simple, each step must be analysed to provide
a clear understanding.
Our sentiment indicator is built in three steps. First, suitable finan-
cial market variables, expressing a clear connection with risk, are
selected. The following variables have been selected as input risk
factors: equity volatility (VIX index), FX volatility (average of G4 3M
volatility), interest rate volatility (average of G4 1m1y and 1y5y
swaptions), credit spreads (iTraxx index), swap spreads (2y, G4
average) and the ratio of gold to gold equity. Second, the scoring
9
technique is developed. Of the six variables selected, a score is

assigned depending on the current value of the variable over the
time horizon. Each day, the variables are sorted based on the last 30
days of data, and are assigned a score of one if they have the highest
value in the past 30 days (extreme risk off) or two if they have the
second highest value, all the way to 30 for the lowest value. Last, a
simple weight of 1/6 is assigned to each of the variables. The average
of the six scores is linearly projected in the interval 01, with
low/high values representing risk aversion (off) if the sentiment
indicator falls below 0.35, risk-seeking (on) sentiment (above 0.7)
and risk neutral (between 0.35 and 0.7). These bands can be seen in
Figure 1.9 and illustrate the strong connection between the Dow
Jones-UBS (DJUBS) commodity index and the sentiment indicator.
In addition to the 30-day sentiment indicator, here we develop a
100-day and a 252-day sentiment indicator for a commercial applica-
tion. The methodology/scoring method is identical, but the
look-back period is 100 days and 252 days, not 30. The reason for a
longer look-up is intuitive. Imagine a scenario where commodity
prices are trending down, say, for 60 days. The 30-day sentiment
indicator has to turn upwards within those 60 days because the
scoring is based on the last 30 days, and so even in a declining market
the sentiment may rise. In this sense, the 30-day sentiment indicator
is a short-term indicator, and we develop the 100-day and 252-day
indicator to assess medium- and longer-term trends. Obviously, with
the 100-day indicator the sentiment is less volatile and would enable
Figure 1.9 SG sentiment indicator and the DJUBS (5d ma) returns: a strong link
1.0 Sentiment Indicator DJUBS (5d ma) 0.015
0.8 Risk seeking 0.01
0.6 0.005
0.4 0
0.2 -0.005
Risk averse
0.0 -0.01
Feb-12 Mar-12 Apr-12 May-12 Jun-12 Jul-12 Aug-12 Sep-12
10
an investor to hold positions for longer (as the investment is based on

sentiment) and incur lower trading costs from rebalancing.
Overlaying with an even longer timeframe (252 days) would add a
further layer of security, ensuring that in periods of extended risk
aversion one does not see a return to risk seeking prematurely,
which may be signaled by a 30-day indicator. Regardless of the look-
back period, the methodology is simple and its relationship to
commodity prices extremely strong. Indeed, it is difficult to find a
daily indicator with such a strong short-term relationship to almost
every commodity within the DJUBS (see below).
SENTIMENT CAUSES COMMODITY PRICES AND NOT THE

OTHER WAY AROUND
Of interest is the question of causality and the speed of response of
the DJUBS to changing sentiment. To this end, we estimated a
reduced-form five lag VAR (vector-auto-regression) using daily
(stationary) data from early 2007 to mid-2012 (technical details
excluded to conserve space). Resulting causality tests confirm at very
high levels of confidence (5%) that sentiment causes DJUBS price
movements, and not the other way around. Here we can take the
causality analysis one step further with the assistance of impulse
response functions. We shock our VAR model by one standard devi-
ation (down) and trace out the influences of sentiment on the DJUBS
price path, and vice versa. Focusing on Figure 1.10, we see that a one
Figure 1.10 Impulse response: a one standard deviation drop in sentiment drags
down the DJUBS to its lowest level after five days
Std. Dev
0.2 Reaction of DJUBS to a drop in sentiment over 10 days
0
1 2 3 4 5 6 7 8 9 10
-0.2
-0.4
-0.6
-0.8
-1
-1.2
11
Figure 1.11 But a one standard deviation drop in the DJUBS does not influence
sentiment
Std. Dev
0.1
0.08
0.06
Reaction of sentiment to a drop in DJUBS over 10 days
0.04
0.02
0
1 2 3 4 5 6 7 8 9 10
standard deviation decline in sentiment results in a negative price

path for DJUBS ie, it also declines. What is interesting, however, is
the speed of that response and the time it takes for DJUBS to fully
incorporate the negative sentiment.
The first day after the shock (day 1) DJUBS prices react, but by day
five, DJUBS has declined by the same amount, in percentage terms,
as the negative sentiment. Beyond day five, DJUBS returns to its pre-
shock level. The equivalent decline in DJUBS prices (one standard
deviation) does not have a significant influence on sentiment (but
raises it modestly) see Figure 1.11.
FOR ALMOST EVERY COMMODITY, 2008 RESULTED IN A

STRUCTURAL SHIFT IN ITS RELATIONSHIP WITH SENTIMENT
Measuring the relationship between variables at various points in
time, rather than using a single correlation coefficient over the entire
sample period, provides information on the evolution of the relation-
ship dynamically. For this purpose, simple correlation measures such
as rolling historical correlations and exponential smoothing are
widely used. The rolling historical correlation estimator provides
equal weights to newer and older observations, and raises issues
surrounding window-length determination. The exponential-
smoothing estimator requires the user to adopt an ad hoc approach to
choosing the smoothing parameter. The dynamic conditional correla-
12
tion (DCC) methodology developed by Engle (2002) helps to remedy

both of these issues.1 In the first step, time-varying variances are esti-
mated using a general autoregressive conditional heteroscedasticity
(Garch) model. In the second step, a time-varying correlation matrix is
estimated using the standardised residuals from the first-stage esti-
mation. Here, we use the DCC method because it has been shown to
outperform other widely used correlation structures in helping with
portfolio investing decisions.2 To assess the relationship between
commodities and sentiment, we correlate the rolling nearby futures
contract prices (using log returns) for each component of the DJUBS
with the 30-day sentiment indicator with daily data beginning in
September 2006. Importantly, the results are qualitatively very similar
when we correlate the DJUBS component prices with the 100-day
indicator (results excluded to conserve space).
Figures 1.12 and 1.13 plot the time-varying correlation of the log
returns of aluminium (LHS) and copper prices (RHS). September 15,
2008 (Lehman bankruptcy) was a game changer there is a notice-
able shift in the relationship between the base metals markets and
sentiment. For aluminium, the average correlation tripled (from 0.13
to 0.38) with the maximum correlation post-Lehman reaching 0.59.
The minimum correlation post-Lehman was 0.15, still higher than
the average pre Lehman. In the case of copper, the correlation
increases from an average of 0.11 to 0.42, almost four times higher
post-2008. Interestingly, the volatility of the correlation of copper
Figure 1.12 DCC between aluminium prices and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
0
5
1
08
11
6
9
7
2
-0
-0
-1
-1
-0
l-0
-0
l-1
n-
n-
ct
ct
ct
l
pr
Ju
-Ju
Ap
-Ju
Ja
Ja
O
-O
-O
-0.1
-A
-
-
12
12
12
-
12
12
12
12
12
12
12
13
Figure 1.13 DCC between copper prices and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
0
6
08
11
1
5
2
7
l-0
l-0
-1
-0
-0
-1
l-1
-0
n-
n-
ct
ct
ct
pr
pr
-Ju
-Ju
-J u
-Ja
-Ja
-O
-O
-O
-0.1
-A
-A
12
12
12
12
12
12
12
12
12
12
Figure 1.14 DCC between Brent and sentiment

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
5
1
0
08
11
6
9
7
2
-0
-0
-1
-1
l-0
l-0
-0
l-1
n-
n-
ct
ct
ct
pr
pr
-0.1
-Ju
-Ju
-Ju
-Ja
-Ja
-O
-O
-O
-A
-A
12
12
12
12
12
12
12
12
12
12
and sentiment is half that of aluminium, post-Lehman. Turning now

to the energy markets, here represented by Brent and heating oil, the
results also illustrate a structural break. Pre-Lehman, the Brent corre-
lation was a mere 0.08, post-Lehman it was 0.39 (see Figure 1.14). For
heating oil (Figure 1.15), we see the correlation rise from an insignifi-
cant 0.04 to 0.38.
At first glance, the results of gold and sentiment may appear coun-
terintuitive, as their average correlation pre-Lehman was 0.06 (see
Figure 1.16). While low, their post-Lehman correlation (relative to
14
Figure 1.15 DCC between heating oil and sentiment

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
6
08
11
8
1
0
2
7
l-0
l-0
-0
-1
r-1
l-1
r-0
n-
n-
-0.1
ct
ct
-Ju
-Ju
-Ju
p
p
-Ja
Ja
-O
-O
-A
-A
-
12
12
12
12
12
12
12
12
12
-0.2
Figure 1.16 DCC between gold and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
-0.1
-0.2
-0.3
5
08
11
2
-0
l-0
r-0
-0
l-0
r-1
-1
l-1
n-
n-
ct
ct
ct
-Ju
-Ju
-Ju
p
p
-Ja
-Ja
-O
-O
-O
-A
-A
12
12
12
12
12
12
12
12
12
12
other markets) is not much higher at 0.18, on average. Interestingly,

for both gold and silver, the Lehman event did increase the correla-
tion, but it was not a structural change, in the way it was for the energy
and base metals markets. However, gold is a unique commodity,
driven as much by sentiment, macro and the dollar as by its own
fundamentals (eg, central bank involvement, exchange-traded fund
15
(ETF) volumes, jewellery and coin demand, mining and scrap

supply), so a decrease (increase) in sentiment may result in an increase
(decrease) in gold demand, hence dragging their correlation lower.
Gold is often negatively related in periods of extreme crisis, hence
fulfilling its role as a flight to safety. Post-Lehman, and after the
August 2011 euro crisis, there has been a negative correlation with
sentiment. This is less evident in silver (Figure 1.17), the more indus-
trial of the two precious metals. Its correlation rose from an average
of 0.16 pre-Lehman to 0.30 post-Lehman.
Not surprisingly, the role of sentiment is not as important to the
agricultural markets despite their reacting to the Lehman crisis in the
same way as the base metals and energy markets (albeit at a much
lower level). The scale of the axis hides the subtle nature of the
change: it was very low in the case of corn (Figure 1.18). From a pre-
Lehman correlation of 0.13, we only see a rise to 0.16. Hardly
significant, for coffee we see a rise from 0.10 to 0.22, a doubling of the
correlation (Figure 1.19).
Not shown (to conserve space) is the change in the relationship for
wheat. The correlation before Lehman was actually negative, on
average; post-Lehman, it averages 0.19. Therefore, agriculture
which was less influenced, certainly in the short run, by the euro
crisis, or by a slowdown in Chinese demand which would influence
the more cyclical commodities as with base metals and energy, is
not going to be as affected by things outside of its own fundamentals.
As we have illustrated, agriculture markets are still positively related
Figure 1.17 DCC between silver and sentiment

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
6
08
11
5
1
7
2
l-0
l-0
-1
-0
-0
-1
r-0
l-1
n-
n-
-0.1
pr
ct
ct
ct
-Ju
-Ju
-Ju
p
-Ja
-Ja
-O
-O
-O
-A
-A
12
12
12
12
12
12
12
12
12
12
-0.2
16
Figure 1.18 DCC between corn and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
0
12-Oct-05
12-Jul-06
12-Apr-07
12-Jan-08
12-Oct-08
12-Jul-09
12-Apr-10
12-Jan-11
12-Oct-11
12-Jul-12
Figure 1.19 DCC between coffee and sentiment

0.35
0.3
0.25
0.2
0.15
0.1
0.05
-0.05
5
08
2
-0
l-0
-0
-0
l-0
-1
-1
l-1
n-
n-
ct
pr
ct
pr
ct
-Ju
-Ju
-Ju
-Ja
-Ja
-O
-O
-O
-A
-A
12
12
12
12
12
12
12
12
12
12
17
to sentiment and are still part of global benchmark indexes, but senti-
ments influence on them is certainly lower.
Last, we present a couple of examples of markets that did not
change after Lehman. US natural gas (a domestic rather than global
market) is the distinct outlier in that its correlation pattern did not
change at all with the structural change in 2008 (see Figure 1.20). Its
average correlation remained at 0.06, precisely the same value it had
before the crisis in 2008. Lean hogs is also independent of sentiment,
having a very similar correlation value pre- and post-Lehman (0.06
and 0.05). Its correlation can occasionally go negative (Figure 1.21).
Figure 1.20 DCC between US natural gas and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
0
5
08
1
7
2
-0
l-0
-0
l-0
-1
-1
r-0
l-1
n-
n-
ct
ct
pr
ct
-0.1
-Ju
-Ju
-Ju
Ap
-Ja
-Ja
-O
-O
-A
12
12
12
-
-
12
12
12
12
12
12
12
-0.2
Figure 1.21 DCC between lean hogs and sentiment

0.6
0.5
0.4
0.3
0.2
0.1
0
5
08
11
1
7
2
-0
-1
l-0
-0
l-0
-1
-0
l-1
n-
n-
-0.1
ct
ct
ct
pr
-Ju
Ju
Ap
-Ju
-Ja
Ja
-O
-O
-A
-
12
12
12
-
-
12
12
12
12
12
12
12
-0.2
-0.3
18
DO SOME COMMODITIES REACT MORE TO SENTIMENT IN

RISK-OFF VERSUS RISK-ON ENVIRONMENTS?
We conclude our analysis of the relationship between commodities
and sentiment by digging deeper into the relationships during
periods of risk off, risk neutral and risk on before the Lehman
crisis (see Table 1.1, left-hand side) and post-Lehman (right-hand
side). First, we present the rankings from the 30-day sentiment indi-
cator in Table 1.1. Prior to the Lehman bankruptcy (left-hand side),
the top 10 commodities most correlated with sentiment in risk off
Table 1.1 Ranking of correlations (20 = least, 1 = most) between components of

the DJUBS and 30-day sentiment (pre-Lehman, January 2006September 2008,
post-Lehman, September 2008September 2012)
Pre-Lehman Post-Lehman
Risk Risk Risk Risk Risk Risk

off neutral on off neutral on
Zinc 1 1 2 Aluminium 1 4 4
Silver 2 4 4 Copper 2 1 1
Cotton 3 1 1 Brent 3 2 2
Copper 4 8 8 Heating oil 4 3 3
Nickel 5 3 3 WTI 5 6 6
Aluminium 6 5 5 RBOB 6 8 8
Gold 7 14 18 Zinc 7 5 5
Corn 8 6 7 Nickel 8 7 7
WTI 9 10 10 Silver 9 9 10
RBOB 10 7 6 Soybean oil 10 10 9
Brent 11 12 11 Soybeans 11 11 11
Coffee 12 9 9 Coffee 12 12 12
Soybean oil 13 18 15 Cotton 13 13 13
Soybeans 14 16 16 Wheat 14 14 14
Sugar 15 17 12 Corn 15 16 16
Natural gas 16 13 14 Gold 16 15 15
Heating oil 17 15 17 Live cattle 17 17 17
Lean hogs 18 11 13 Sugar 18 18 18
Live cattle 19 20 20 Natural gas 19 19 19
Wheat 20 19 19 Lean hogs 20 20 20
19
were represented by all commodity types: base metals, precious

metals, agriculture and energy. In fact, energy only just makes the
top 10, with West Texas Intermediate (WTI) ranked ninth in terms of
its correlation with sentiment in risk-off environments.
When we focus on the post-Lehman period, the patterns change
considerably. The top 10 most-correlated commodities with senti-
ment are base metals and energy and silver (which one could argue
is somewhat of an industrial metal). There are no more agricultural
commodities at the top (with risk off) until we get to number 10:
soybean oil (which moves into number nine (just) in risk-on envi-
ronments). Moreover, regardless of the environment, risk on, risk
neutral or risk off, the rankings of commodities hardly change
post-Lehman. The most significant change is aluminium, which
moves from being the most correlated with sentiment in risk off to
being the fourth most correlated in risk off: a relatively minor
change. Compare this to gold, for example, pre-Lehman. Its ranking
changes from the seventh most correlated with sentiment in risk-
off to 18th in risk-on environments. The bottom line is, with
changes in sentiment, base metals and energy are much more influ-
enced by sentiment than other types of commodities post-Lehman.
BRINGING IT TOGETHER: A SIMPLE OVERLAY EXAMPLE TO

THE DJUBS
In this section, we will illustrate how to incorporate the main results
from our research into a simple product for investors wishing to
benchmark against the basic DJUBS (excess return) long-only expo-
sure. There are obviously numerous applications, but for clarity and
simplicity we focus on a simple overlay. We simply try to incorpo-
rate a medium-run sentiment indicator (100-day) to help with
re-weighting overlaid with a longer-term indicator (252-day) to
provide a further layer of insurance in periods where prices fall for a
long period of time. Critically, this is just an example and many other
applications can be made. Here is the procedure.
First, we develop two sentiment indicators based on the princi-

ples outlined in the previous section. One is the sentiment
indicator based on the 100-day look-up period to signal re-
weighting decisions (to reduce trading costs that occurs to
shorter-run indicators). The second sentiment indicator has a
20
252-day look-back period. The 252-day sentiment indicator is

going to behave differently than the 100-day indicator as it incor-
porates a much longer-term risk appetite. Therefore, the 252day
indicator provides insurance for periods of extended
declines/risk aversion that even the 100-day indicator might not
pick up on.
Second, using the 100-day indicator, we focus on weight tilting
ave already identi-
g natural gas) the
nt nine) have the
-off and risk-on
entiment indicator
ff is where senti-
ween 35% and 70%
e start in the risk-
invested fully into
on the close of the
o fully account for
Then, on a day that
vironment, we re-
modities and into
e preserve the rela-
JUBS but distribute
ther commodities
ties weights equal
nts, we reduce the
zero and re-weight
ning their relative
a longer-term view
h a risk switch that
52-day sentiment is
or falls below 20%
e set to zero (again,
roducts (alpha), for
his is an arbitrary
nvestors risk toler-
s is that, if there is
ue to a crisis (for
21
example), the investor sits on the sidelines until long-term senti-

ment improves (ie, passes through 20% to the upside), at which
point the risk-tilting mechanism kicks in once more.
Importantly, while we have isolated the commodities most affected

by sentiment as the ones to remove in risk off versus capture in
risk on, even the less-sensitive commodities are positively corre-
lated with sentiment. Therefore, one could argue to completely
remove all commodity exposure in risk off, but here we choose to
remain invested instead of having extended periods of time sitting
on the sidelines. Performance is clearly better overlaying the senti-
ment (see Figure 1.22), a function of re-weighting and overlaying
with an extra layer of insurance (the 252-day window). Focusing
purely on the September 2008 onwards (47 months), the number of
positive months increases from 27 to 32 and average annualised
return rises from 13.34% investing in the DJUBS (about 3.52%
annualised) to 74.08% (about 15% average annualised) investing in
the DJUBS, weight-tilted 100-day sentiment indicator with the 252-
day overlay. Ignoring the 252-day overlay (from sentiment) results in
a return of 32%. Therefore, most reward from using the sentiment
indicator comes from the performance attributed from shifting
weights based on the 100-day indicator (about 45% over the DJUBS),
although the overlay (insurance) adds almost the same amount. The
number of times the portfolio is re-weighted because of a change in
sentiment is approximately 25 times per annum. With the 30-day
sentiment indicator applied (instead of the 100-day), the number of
times is 46 hence higher trading costs.
SUMMARY
It is clear that outside influences on commodities have picked up
since 2008. The role of macro, dollar and liquidity vary across
commodities and across time. Sentiment has made a substantial
impact on the commodities markets since 2008. Here, we have docu-
mented the causal relationship (from sentiment to commodities) and
reported that some commodities are more affected by sentiment than
others. A ranking was established. We applied our research results
by overlaying the DJUBS with the sentiment indicator signals, util-
ising the rankings of the sensitive commodities by re-weighting in
risk-off and risk-on environments. The re-weighting alone
22
Figure 1.22 DJUBS versus DJUBS-with-weight-tilt (based on 100-day sentiment) and 252-day sentiment indicator overlay
200
180 Incremental return since 2008: overlaying with 252 day Sentiment Indicator: 42%

160
140
120
100
80
Overlaying with sentiment was
60 actually detrimental before 2008
Incremental return since 2008: applying weight tilts to DJUBS: 45%
40
20 DJUBS DJUBS + weight tilt DJUBS + weight tilt + 252 day overlay
0
Sep -06 Sep -07 Sep -08 Sep -09 Sep -10 Sep -11 Sep -12

23
significantly outperforms the long-only DJUBS exposure since 2008

we achieve an extra 45% higher return over the period. However,
overlaying with an extra layer of protection (a signal from a 252-day
sentiment indicator) significantly protects returns from large
declines in commodity prices this adds an additional 42% on top of
the 45%. Total returns using weight tilts and 252-day overlay equals
74.08% since 2008, compared to 13.34% by investing the DJUBS.
The purpose of this chapter is not to suggest fundamentals do not
matter they do, but what is clear is that an analysis of commodity
markets requires something more than counting barrels or bushels.
Even basic applications of sentiment onto commodity markets add
outperformance and significant protection.
1 Engle, R., 2002, Dynamic Conditional Correlation A Simple Class of Multivariate

GARCH Models, Journal of Business and Economic Statistics, 20(3), pp 33950.
2 Huang, J. Z and Z. Zhong, 2010, Time Variation in Diversification Benefits of Commodity,
REITs, and TIPS, working paper, Department of Finance, Pennsylvania State University.
24
2
The North American Natural
Gas Market
Stinson Gibner
Whiteside Energy
This chapter will provide an overview of the most important supply

and demand developments for natural gas, beginning with a brief
discussion on natural gas and how it is traded. The analysis of gas
demand fundamentals and gas production leads to an under-
standing of the dynamics of the storage market for natural gas. The
geographic distribution of sources and demand for gas will also be
examined, before we move on to price dynamics, aided by examples
of how many of these factors influence market prices for gas. In
conclusion, key factors that will determine the future evolution of
prices are identified.
OVERVIEW
What makes the North American natural gas market unique? The
most important factor is that it is a self-contained system within the
confines of North America, apart from limited liquefied natural gas
(LNG) import and export capability. Consequently, the market can
by analysed by understanding supply, demand and storage stocks
within the US and Canada. LNG imports can be relevant, but having
been at less than 2% of the annual supply for many years, they have
little market influence.
Highly seasonal demand driven by winter heating and a lesser
peak from summer cooling loads combines with relatively constant
production flows to require massive storage facilities that can inject
gas during times when supply outpaces demand and withdraw
25
when gas burn rises. Injections and withdrawals from storage facili-
ties are surveyed and reported by the Energy Information
Administration (EIA), part of the US Department of Energy (DOE),
providing a closely watched weekly monitor of supply/demand
balance.
The natural gas transported through long-haul pipelines is
primarily methane with a mixture of some ethane and smaller
amounts of heavier hydrocarbon gases, and may contain a small
percentage mixture of nitrogen and carbon dioxide. The average
heating value of gas consumed in the US is now about 1,025 Btu per
cubic foot or 1.025 million Btu (MMBtu) per thousand cubic feet
(Mcf). This leads to an often-used rule of thumb conversion factor
that 1 Mcf approximately equals 1 MMBtu. Pipelines have specifica-
tions for the range of gas quality acceptable for receipt. The heating
value of the gas accepted must typically lie within a range of, for
example, ~9701,100 Btu per cubic foot. Some of the most common
natural gas units of measure and conversions are given in Table 2.1.
GAS MARKETS
Before the 1990s, natural gas purchases and sales were predomi-
nantly handled by long-term contracts for physical natural gas.
Natural gas can still be traded by the purchase or sale of physical gas
where the seller delivers and the buyer receives the molecules, and
there is also a liquid market where gas can be traded purely finan-
Table 2.1 Common units of measure and conversions
Common units of measure
MMBtu Million Btu

Mcf Thousand cubic feet
Bcf Billion cubic feet (1,000 Mcf)
Tcf Trillion cubic feet (1,000 Bcf)
Bcm Billion cubic meters
MMT Million tonnes
MMBOE Million barrels of oil equivalent
Conversions
1 Bcm = 35.3 Bcf

1 MMT of LNG = 48.7 Bcf methane
1 MMT of LNG = 1.38 Bcm
1 MMBOE ~ 5.6 Bcf (conversion varies)
26
THE NORTH AMERICAN NATURAL GAS MARKET
cially. Most financial market instruments derive from the traded

structures in the physical gas market.
The physical gas market traditionally trades gas for both monthly
and next day delivery. Purchases of monthly gas are for gas to be
delivered in approximately equal daily quantities over an entire
calendar month. The majority of these physical gas purchases and
sales are made during bid week, the last week of each month. Gas
also trades in the daily market, with purchases and sales of gas typi-
cally occurring during the morning hours prior to the gas flow date
in order to allow time for proper nominations for gas flows on the
required delivery pipelines. Gas for the weekend and Monday are
traded on the preceding Friday.
The Nymex natural gas futures contract was introduced in 1990,
and it rapidly grew in traded volumes. The contract can be physi-
cally settled at the Henry Hub in southern Louisiana, which allows
for the interchange of gas between 13 pipelines, or at an alternate
delivery point based on mutual agreement between the buyer and
seller. Monthly futures contracts are listed, each contract unit repre-
senting 10,000 MMBtu, with the contract price quoted in US$/
MMBtu and having a tick size of US$0.001 (0.1 cent) per MMBtu.
Although many months of futures are listed, liquidity concentrates
at the front of the futures curve. In addition to trading on the Chicago
Mercantile Exchange (CME), Henry Hub futures are listed on the
IntercontinentalExchange (ICE).
Of course, gas trades, both physical and financial, for many
delivery locations throughout North America other than Henry
Hub. In order to facilitate these transactions, a large number of gas
price indexes have been created. The primary publishers of these
indexes are Platts and Natural Gas Intelligence. Each of these
publishers conduct daily and monthly polls of market participants in
order to estimate a representative market price transacted for natural
gas at a variety of geographical delivery areas. The published gas
indexes allow for managing a financial exposure to the gas index
price without transacting any physical natural gas. For example, July
gas at Chicago Citygate may be trading in the forward market for
US$5.00/MMBtu, and a financial buyer may enter a contract to pay
US$5.00 and receive the Chicago Citigate index price after it is
published at the beginning of July.
27
DEMAND SIDE DYNAMICS FOR NATURAL GAS

Most consumption of natural gas falls into four of the categories used
by the EIA: residential, commercial, industrial and electric genera-
tion. Residential and commercial use is primarily for heating, and
both sectors are characterised by a strong winter demand peak and
very flat demand in the summer. Industrial use has much less
seasonality, but about 10% does go toward heating demand in
winter. Electric generation burn peaks in the summer, when air
conditioning loads are the highest. In 2011, residential plus commer-
cial users consumed 32%, power generation 31% and industrial users
28% of all gas consumed in the US.
Industrial use
Industrials use gas for space heating, process heat and also as a feed-
stock. As can be seen in Figure 2.1, industrial demand in the US
decreased dramatically from 1997, dropping by a total of almost 5.5
Bcf/day before bottoming in 2006. Since then, industrial use has
rallied by more than a Bcf per day, interrupted by the Great
Recession year of 2009.
Figure 2.2 deconstructs industry demand by sector; we find that,
Figure 2.1 Average annual industrial gas use (Bcf/day)

24 US$40.00
Industrial use (Bcf/day)
23 US$36.00
Henry hub annual avg. spot price (right axis)
22 US$32.00
Industrial demand (Bcf/day)
21 US$28.00
20 US$24.00
19 US$20.00
18 US$16.00
17 US$12.00
16 US$8.00
15 US$4.00
14 US$0.00
98
99
00
03
05
06
08
09
10
11
97
02
04
07
12
0
19
19
20
20
20
20
20
20
20
20
20
19
20
20
20
20
Source: EIA
28
from 1998 to 2006, there was declining use in every significant sector
except food and non-metallic minerals. The largest losses in use were
in chemicals manufacturing (down 2.59 Bcf/day), primary metals
(down 0.82 Bcf/day) and refining (down 0.43 Bcf/day). Within the
chemical sector, nitrogenous fertilisers alone accounted for almost
0.75 Bcf/day loss of demand over this time period due to production
moving offshore. Imports of anhydrous ammonia grew by 4.4 million
short tons, equating to 0.45 Bcf/day of domestic gas demand loss.
Since 2006, industrial use of gas has begun to grow again. Of
course, the deep recession between late 2008 and early 2010 created a
loss of demand of around 1.5 Bcf/day in 2009. However, growth of
industrial demand has started to accelerate due to low natural gas
prices, which looks to continue into the future, driven by a resur-
gence in the chemical and refining sectors. Domestic ammonia
Figure 2.2 Largest industrial consumers of natural gas (Bcf/day)
8.0
1998
7.0
2002
2006
6.0 2010
5.0
Bcf/day
4.0
3.0
2.0
1.0
0.0
s
ls
s
pe
al
ie
al
al
in
a
Fo
ic
et
or
er
et
fin
Pa
em
m
in
g
re
te
m
y
ed
Ch
ca
um
ar
at
im
li
er
ic
le
al
th
Pr
br
tro
et
O
Fa
Pe
on
N
Source: EIA
29
production has been stepped up again, and a number of corporations

have announced plans to build new chemical plants to take advan-
tage of the low energy prices in the US. There have even been
announcements of new metal-processing plants to be built,
expanded, or reopened. It appears likely that 2013 industrial
consumption will be at least 2 Bcf/day above the levels of 2006, and
growth should continue to be robust for a number of years as new
use facilities come online.
Power generation
Because power generation is a large and growing source of demand
for natural gas, an understanding of the power markets is critical in
anticipating future levels of gas demand. Increasing use of gas for
power generation has provided the largest increase of any sector.
Figure 2.3 shows monthly average gas burn for power generation
and the upward trend in demand since the early 2000s. Figure 2.4
shows that this steady growth in gas burn for generation continued
even through years of little or no growth in total power demand.
This trend is poised to continue as the phasing in of air pollution
standards for coal plants leads to continued coal plant retirements.
Figure 2.3 shows that monthly gas burn also comprises strong
seasonality of gas generation burn with the distinct summer air
conditioning demand peak and the much smaller winter heating
demand peak that has emerged.
Most of the growth in gas burn for power since the early 2000s has
come at the expense of decreasing coal-fired generation. Figure 2.5
shows the annual mix of generation sources for the 11-year period
ending in 2012. During this time, the percentage of generation from
nuclear plants and from sources other than coal, gas and nuclear
(which leaves hydroelectric, other renewables and liquid fuels) has
held roughly constant, so there has been an almost one-to-one trade-
off in loss of coal generation with gain in gas generation. Gas
generation has grown from 17.9% in 2002 to 30.4% of total US gener-
ation in 2012, while coal has fallen from 50.1% to 37.4% over that
time. We should note that 2012 was an exceptionally high year for
gas burn due to conditions that may not recur in the near future.
In fact, power generation provides one of the few demand sectors
that can significantly change the fuel mix based on short-term fuel
price levels and economics. During the period of cheap oil in the
30
31
0
5
10
15
20
25
30
35
Ja
n-
20
01
Source: EIA
Ju
l-2
00
Ja 1
n-
20
02
Ju
l -2
00
2
Ja
n-
2 00
3
Ju
l-2
00
Ja 3
n -2
0
04
Ju
l-2
00
4
Ja
n-
20
05
Ju
l-2
00
Ja 5
n-
20
06
Ju
l-2
00
Figure 2.3 Monthly average gas use for electric generation (Bcf/day)
Ja 6
n-
20
07
Ju
l-2
00
7
Ja
n-
20
08
Ju
l-2
00
Ja 8
n-
20
09
Ju
l-2
00
Ja 9
n-
20
10
Ju
l-2
01
Ja 0
n-
20
11
Ju
l-2
01
Ja 1
n-
20
12

Figure 2.4 US annual power generation (million GW hours)

4.20
4.15
4.10
Million gigawatthours
4.05
4.00
3.95
3.90
3.85
3.80
3.75
3.70
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: EIA
Figure 2.5 Percentage of annual power generation by energy source

60.0%
Coal
50.0%
40.0% Natural gas
30.0% Nuclear
20.0%
All other
10.0%
0.0%
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: EIA
1990s and early 2000s, fuel oil was sometimes economically competi-
tive with natural gas, so during times of high gas prices there could
be an economic incentive to turn on oil-fired generation which, in
turn, liberated gas for higher value heating use. With the advent of
oil prices near US$100+/bbl, natural gas has remained much less
expensive and oil use for generation has fallen from the already low
level of 2% of total generation in 2002 to 0.3% in 2012.
32
Latterly, coal-to-gas substitution has become a key factor to watch

for understanding demand trends for natural gas. The relative costs
of generating power from coal and gas drive substitution economics.
To calculate the cost of generation, we must know how much coal or
gas it takes to generate a megawatt (MW) of power. The amount of
fuel required per unit of power generated is called the heat rate. For
actual generation plants, the heat rate will depend on a number of
factors including type of equipment, generation level (% of
maximum capacity) and ambient air temperature. After estimating
the heat rate, fuel cost and variable operating and maintenance cost,
the marginal cost of power production can be calculated for a plant.
Many analysts construct stack models, in which plants are
stacked in order of their production costs, then the markets marginal
cost of production can be found for a given level of net power
demand, and the amount of expected gas burn and coal burn can be
calculated. Of course, there are many additional details involved in
this process, including estimation of load served by nuclear and
renewable sources, forecast of power imports and exports to
connected regions, plant maintenance and forced outage rates, and
the influence of operational optimisations to minimise start costs. In
practice, stack models are difficult to calibrate for accurately fore-
casting future market prices, but they can be quite useful in more
qualitative analysis of market trends and behaviour.
Figure 2.6 tells of an interesting chapter in the natural gas demand
growth story. US power load growth accelerated in the mid-1990s at
the same time that uncertainties about market deregulation and
about future coal plant environmental regulations led to a reluctance
to build additional coal-fired generation. The market reacted by
beginning an unprecedented build of new gas-fired generation,
which can be seen by the huge increases in gas capacity as new plants
came online in 2002 and 2003. The build rate slowed but has
continued through the last decade. In addition to making more gas-
fired generation capacity available, the new and more efficient plants
have lowered the average heat rate of the available gas-fired genera-
tion fleet. The average heat rate of gas generation has dropped from
just over 10 MMBtu/MWh in 2001 to 8.15 MMBtu/MWh in 2011,
and the US continues to build new combined cycle gas turbines with
heat rates near 7 MMBtu/MWh.
The shift away from coal toward gas generation is set to continue,
33
Figure 2.6 Natural gas and coal generation capacity and gas average heat rate
450 12.0
NG summer capacity (GW)
Gas average heat rate (mmbtu/MWh)

11.5
Coal summer capacity (GW)
400 Average heat rate 11.0
Capacity (gigawatts)
10.5
350 10.0
9.5
300 9.0
8.5
250 8.0
7.5
200 7.0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: EIA
with over 30 gigawatts (GW) of additional coal plant retirements

planned between 2013 and 2018. In that time period, combined cycle
gas generation capacity may grow by almost 60 MW if all planned
units are permitted and built.
Residential and commercial demand

While the residential and commercial (rescom) use of gas has not
displayed the growth seen in the generation sector, there are substan-
tial year-to-year variations in total use. The largest driver of demand
variability in rescom use are winter temperatures, which influence the
amount of gas needed for home and commercial heating during the
cold months of the year. As can be seen in Figure 2.7, there has been
considerable variability in the weather-sensitive heating demand
months, but no obvious trend or much change in summer demand
levels since the early 2000s. This suggests that growth in the number of
consumers has been offset by conservation and heating efficiency
gains, resulting in very little (if any) net demand growth.
Exports
The US plans to begin exporting LNG from the Gulf Coast. Sabine
Pass LNG facilities target around early 2016 for beginning LNG
exports. With US gas prices likely to remain in the range of
US$4.006.00 MMBtu, landed prices to Europe would likely be in the
range of US$8.0011.00 MMBtu. Export volumes are expected to
34
35
Bcf / Day
Ja
n-
0
10
20
30
40
50
60
20
01
Source: EIA
Ju
l -2
00
Ja 1
n-
20
02
Ju
l-2
00
2
Ja
n-
20
03
Ju
l-2
00
Ja 3
n-
20
04
Ju
l-2
00
4
Ja
n-
20
05
Ju
l-2
00
Ja 5
n-
20
06
Ju
l-2
00
Ja 6
n-
20
07
Ju
l-2
00
7
Ja
n-
Figure 2.7 Residential and commercial gas demand (monthly average in Bcf/day)
20
08
Ju
l-2
00
Ja 8
n-
20
09
Ju
l-2
00
Ja 9
n-
20
10
Ju
l-2
01
Ja 0
n-
20
11
Ju
l-2
01
Ja 1
n-
20
12

reach over 1 Bcf/day in 2016 and planned projects would grow

exports to over 3 Bcf/day by 2018, suggesting that the EIAs
projected 201320 total production growth of about 5 Bcf/d (shown
in Figure 2.11) may be low compared to the likely demand growth.
SUPPLY SIDE CONSIDERATIONS

The US meets its gas needs primarily with domestic production and
imported gas from Canada. LNG imported by tanker from overseas
locations provides a third source of supply. Figure 2.8 shows histor-
ical monthly production since 1993. Production grew slowly in the
1990s and peaked in March of 2001. Production then began a series of
annual declines that led many to believe that domestic US gas supply
might be permanently headed in that direction. LNG imports were
seen as the solution to securing additional gas supply. In 2000, the US
had two operating LNG import facilities: Everett and Lake Charles.
Two additional existing facilities, Elba Island and Cove Point, moth-
balled in the early 1980s, were re-commissioned and began receiving
deliveries in 2001 and 2003, respectively. In addition, the Federal
Energy Regulatory Commission (FERC) granted authorisations for
several additional import terminals that were completed and
commissioned in 200811. However, most of these new facilities
have not yet seen heavy use due to the strong resurgence in domestic
production that began in 2007.
Shale gas
The driver of this reversal in fortune for natural gas production was
a combination of new technologies and higher natural gas prices,
which allowed shale gas to be produced economically in high quan-
tities. Conventional gas production came largely from gas trapped in
sandstone formations with high porosity and permeability, allowing
the gas to flow through the formation to the wellbore. It had long
been recognised that natural gas was also trapped in many shale
formations, but shale is characterised by much lower porosity and
permeability that limits the movement of the trapped gas. Mitchell
Energy began to experiment with a combination of horizontal
drilling and hydraulic fracturing to produce gas from the north
Texas Barnett Shale. After Devon acquired Mitchell in 2002, the
Barnett drilling programme accelerated and, by 2007, the Barnett
Shale produced 1.1 Tcf of gas equivalents making it the second-
36
37
(Bcf / day)
Ja
n-
45
50
55
60
65
70
75
19
93
Source: EIA
Ja
n-
19
94
Ja
n-
19
95
Ja
n-
19
96
Ja
n-
19
97
Ja
n-
19
98
Ja
n-
19
Figure 2.8 US domestic production (Bcf/day)
99
Ja
n-
20
00
Ja
n-
20
01
Ja
n-
20
02
Ja
n-
20
03
Ja
n-
20
04
Ja
n-
20
05
Ja
n-
20
06
Ja
n-
20
07
Ja
n-
Katrina & Rita
20
08
Ja
n-
20
09
Ja
n-
20
10
Ja
n-
Gustav & Ike
20
11
Ja
n-
20
12
TX Cold
Ja
n-
20
13

largest producing field in the US (see Joel Parshall, 2008, Barnett

Shale Showcases Tight-Gas Development, JPT, September). After
this success in the Barnett, many shale fields began to contribute
significantly to US production, and Fayetteville, Haynesville,
Marcellus, Bakken and Eagle Ford all became well-known names in
the oil and gas E&P sector. Shale gas grew from less than 3% of US
gas production in 2003 to more than 40% at the beginning of 2013.
Figure 2.9 shows this growth in production from shale gas fields.
Figure 2.10 shows that the number of drilling rigs directed
towards natural gas production more than doubled from ~700 in
2003 to a peak of almost 1,600 near the beginning of the financial
crisis and recession of 200809. Then, in spite of the gas-directed rig
count plunging back to the 7001,000 levels, natural gas production
continued to grow as shale gas production growth accelerated in
2010 and 2011.
The continued growth in production even with lower gas directed
rig counts can be attributed to a combination of factors, including the
shift of drilling towards horizontal shale wells, improvements in
drilling efficiency and growth in associated natural gas production.
From September 2008 to September 2010, the number of gas-directed
rigs fell from roughly 1,600 to 1,000; however, the number of hori-
Figure 2.9 US shale gas production (Bcf/day)

30
Other US shale gas
25 Bakken (ND)
Eagle Ford (TX)
20 Marcellus (PA and WV)

Haynesville (LA and TX)
Bcf/day
15 Woodford (OK)
Fayetteville (AR)
10 Barnett (TX)
Antrim (MI, IN, and OH)
5
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Source: EIA
38
Figure 2.10 Count of rigs drilling for oil and gas in the US
16 1600
Natural gas price

14
Oil directed rigs
Natural gas directed rigs
12 1200
10
US$/mmbtu
Rigs
8 800

4 400
0 0
95
96
00
01
05
06
10
11
94
97
98
99
04
07
09
02
03
13
-0
-1
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n-
n
n
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Ja
Source: EIA
39
Figure 2.11 Annual US gas production by source
90
Shale gas
Forecast
80 Tight gas
Non-associated offshore
70
Coalbed methane
Associated with oil
60
Non-associated onshore
Gas production (Bcf/day)
50
40
30
20
10
0
1990 1995 2000 2005 2010 2015 2020 2025 2030
Source: EIA, Annual Energy Outlook 2013, early release
40
zontal drilling rigs directed towards gas actually increased from

~500 to about 650 over the same period. In other words, horizontal
drilling grew from one-third of gas rigs to almost two-thirds by late
2010. The much higher average initial production rates from hori-
zontal wells allowed continued production growth with lower rig
counts. At the same time, drillers were learning and improving the
efficiency of their shale-drilling operations, leading to shorter
drilling time and more wells drilled by each active rig, a trend which
continues.
The rapid deployment of oil-directed drilling rigs beginning in
July 2009 can be clearly seen in Figure 2.10. According to the EIA,
natural gas associated with oil was about one Bcf/day higher in 2012
than in 2010, thus adding to natural gas production growth. It should
also be noted that the distinction between drilling categorised as oil-
directed as compared to gas-directed is somewhat imprecise.
Additionally, new natural gas production lags drilling activity,
especially in the new shale production fields, because wells often
must wait for infrastructure to catch up with drilling whereas oil
production can, if necessary, be moved by truck or rail. The only
economically feasible way to move natural gas production from the
wellhead is by pipeline. Therefore, new fields must wait for the
requisite gathering pipeline systems to be constructed to deliver gas
to users and to the long-haul pipeline system. In addition, wet or
sour gas production may need to wait for processing facilities that
remove liquids and impurities before the gas can be delivered to a
major pipeline.
Robust production growth plus the warm winter of 2011/2012 led
to a supply surplus, driving prices down to below US$2.00 for the
first time in years. Gas-directed rig counts plummeted to near 400
rigs, the lowest level in a decade, as many shale gas fields became
uneconomic at the low price levels.
In the long run (but hopefully before we are all dead), one would
expect that natural gas prices should gravitate towards a price level
that makes marginal production economic. However, limited trans-
parency of drilling costs and uncertainties in well production
profiles and estimated ultimate recoveries (EUR) make estimating
production costs difficult. Also, costs and efficiencies change contin-
ually, making drilling economics a moving target. In addition, the
proportion of associated liquid hydrocarbons influences the overall
41
economics as the liquids sell at a premium to natural gas.

Conventional wisdom recognises the Marcellus shale as the lowest
cost of the gas shales, with production costs below US$3.00/MMBtu
in the prime locations. With strong crude oil prices, gas-drilling
returns may have to compete with oil-drilling economics when
exploration and production budgets are decided.
Weather impacts on supply

Certain types of weather events can influence production as well as
demand. The clearly noticeable production drops in August and
September 2005 and September 2008 were caused by hurricanes in
the Gulf of Mexico, where there is substantial offshore gas produc-
tion. Hurricanes Katrina and Rita were both Category 5 storms as
they crossed the production area in 2005, and hurricanes Gustav and
Ike were both Category 4 storms. Smaller hurricanes, and even trop-
ical storms, may cause some disruption to supply as personnel are
evacuated from the storm path and some production platforms may
be shut-in as a precautionary measure. Rita and Katrina shut-in
almost 520 Bcf of production, and the 2008 storms caused a loss of
about 340 Bcf of production. Offshore gas production has been in
decline but remains above 4 Bcf per day. Because hurricanes need
very warm water temperatures to power them, the Gulf hurricane
season runs JuneNovember, with August, September and October
being the most active months.
The production decline seen for February 2011 in Figure 2.11
resulted from very cold temperatures in Texas and nearby states. Gas
production declined from wells freezing off and from conditions that
hampered the ability of pumpers to maintain production. Severely
cold temperatures happen rarely enough in these production areas
that many wells do not have protection against cold temperatures,
allowing water vapour in the natural gas stream to freeze and
constrict flow from the wells. Thus, when unusually cold weather
invades southern and southwestern production areas, freeze-offs are
a danger to production.
Ethane rejection
NGLs, which are comprised of ethane, propane, butane and heavier
hydrocarbons, enhance production value when stripped from the
natural gas stream and sold separately. The stripping of wet gas,
42
carried out by fractionation facilities, may also be necessary to bring

the liquid content of the gas down to standards required by
pipelines. For example, 1.25 MMBtu/Mcf gas may yield around 0.12
bbl of liquids per Mcf. At an average liquids price of US$25.00/bbl,
the liquids alone are worth US$3.00/Mcf and may comprise nearly
half of the value of production.
The lowest value liquid in this stream, ethane, may fall below the
value received by leaving it in the delivered gas stream. In these
cases, the ethane can be rejected during the fractionation process and
effectively increases the net amount of delivered natural gas. That is,
when we say that ethane is rejected, we mean that it is left in the gas
stream with the methane. Ideally, economics will dictate the ethane
rejection decision; however, with the rapid growth of new gas
production in some regions, the infrastructure is sometimes not
sufficient to process all of the produced gas. The total amount of
ethane being extracted from the US gas stream had a heating equiva-
lent value of about 3 Bcf/day of gas in late 2012, and the historical
levels of ethane extraction suggest that varying ethane rejection
could impact net gas deliverability by up to 0.51 Bcf/day.
STORAGE
There is a mismatch between highly seasonal demand as compared
to production which, in the absence of disruptions, trends more
slowly over the years. The large seasonal variability of demand
requires gas to be stored in the low-demand months and withdrawn
in times of high demand. There are over 400 natural gas storage facil-
ities in the US to support this balancing need. Most use depleted gas
reservoirs as the storage space, but leached out underground salt
domes provide almost 8% of the storage capacity and another 8% is
provided by aquifer storage. Reservoirs take many months to fill and
so can be cycled only once a year, although there is usually some flex-
ibility in scheduling the injections and more flexibility in the timing
of withdrawals. Salt domes require much less time to fill, perhaps
one month or less, and so can be cycled many times per year if there
is an economic opportunity to do so. During the 2010/2011 heating
season, a net amount of about 2,200 Bcf was withdrawn from US
storage and then about the same net amount injected during the
summer; however, gross injections plus withdrawals for the year ran
well above the annual net injections plus net withdrawals, showing
43
that many short-term storage injections and withdrawals are made

to support the daily physical market balancing, as well as the annual
seasonal cycle of demand.
Gas storage nomenclature denotes working gas capacity as the
amount of storage gas that can be cycled in and out of storage facili-
ties as part of normal operations. An amount of base gas must be
maintained in the storage facility at all times to maintain the integrity
of the facility. Base gas plus working gas added give the total storage
capacity. Most analysts of supply and demand are mainly interested
in watching the level of working gas in storage, as this represents the
gas available to withdraw for market needs.
As of early 2013, the EIA estimated that US facilities have the
ability to store 4,558 Bcf of working gas. However, the most working
gas actually in storage at any one time was 3,929 Bcf, in autumn 2012.
The EIA also calculates the demonstrated peak working gas
capacity by adding the non-coincident maximums for each facility
to get 4.24 Tcf, 94% of the design capacity. Latterly, additional
storage has been added at a rate of around 75 Bcf of working gas each
year. The maximum working gas capacity becomes quite relevant to
the market in years such as 2012, when the market was oversupplied
and excess production needed to find a home. In spring and early
summer 2012, prices collapsed on fears that storage might fill
completely, but low prices solved the problem as power producers
turned off coal plants and turned on combined cycle gas turbine
(CCGT) plants to burn the inexpensive gas.
Analysts speculate on what minimum amount of working gas the
market requires at the end of the injection season. As can be seen in
Figure 2.12, end-of-season fills since the early 2000s have ranged
from just under 3.2 Tcf to just over 3.9 Tcf. Because of the growth in
use, many believe that the market will now want to be near the high
end of this range to ensure winter reliability of supply.
Each Thursday, the EIA releases a weekly report giving their esti-
mate of the amount working gas in US storage as of the previous
week. This widely anticipated publication gives the single most
important short-term data point about the current supply and
demand balance, and often incites a strong price response from the
natural gas markets. Because of the high importance of the reported
number, fundamental analysts labour daily to forecast it. The EIAs
number itself is based on a statistical model that they use to extrapo-
44
Figure 2.12 Working gas in storage (Bcf)
4,500
4,000
3,500
3,000
2,500
Bcf
2,000
1,500

Range 2004-2010
1,000
2011
2012
500
-
Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Source: EIA
45
Table 2.2 US production and estimated net exports, demand and imports
Production Exports Demand Imports

State or region (Bcf/d)
Texas 20 11
Louisiana 8 5
Oklahoma 6 4
Gulf of Mexico, Federal Offshore 4 4
Arkansas 3 2
Rockies (NM, CO, UT, WY) 15 13
Marcellus (Northeast States) 7 12 4
Midwestern States 1 12 11
California 1 6 5
Florida 3 3
Southeast 1 7 6
late from their population of storage survey respondents to a total US

storage amount, and so has some level of uncertainty itself. This
number represents the net injection or withdrawal summed over all
US storage facilities. Net injections typically begin in late March or
early April, making March the last month of net injections and April
the first month of the year with net withdrawals, except in extreme
conditions such as the warm March of 2012, which left that month
with net injections. During autumn, November is usually the first
month to see weekly withdrawals, although there have been net
withdrawals as early as the last week of October or as late as the first
week of December. In 2006, summer gas demand for power genera-
tion was sufficiently strong and production low enough that there
were net withdrawal weeks in late July and early August.
Because of this seasonality of injections and withdrawals, the
natural gas year is divided into the summer (injection) months of
AprilOctober, and the winter (withdrawal) months of November
March. This seasonality of storage manifests itself in the gas markets
as well. The volatility of the price spread between the October and
November contracts, and the volatility of the price spread between the
March and April futures, are often the highest of all the sequential
month spreads. Also, the term structure of options volatility typically
has local maxima for options on the October and March contracts.
46
GEOGRAPHY OF PRODUCTION AND DEMAND

A large portion of the US gas supply has come from the Gulf Coast
and mid-continent. Texas has the largest gas production at about 20
Bcf/day. Neighbouring Louisiana produces ~8 Bcf/day, and gas
from the Gulf of Mexico Federal Offshore areas comes ashore to
pipelines in Texas, Louisiana and Alabama, and adds another ~4
Bcf/day of supply, although this is less than half of the offshore
supply levels seen in the early 1990s. Additional supply comes from
Oklahoma (~6 Bcf/day) and Arkansas (~3 Bcf/day). There are two
other large supply areas outside of the Gulf coast/mid-continent.
The Rocky Mountain states of New Mexico, Colorado, Utah and
Wyoming combine to produce about 15 Bcf/day of gas, and
Marcellus Shale and other production in Pennsylvania and nearby
states adds about 12 Bcf/day.
Texas, Louisiana and Oklahoma also consume large amounts of gas
for industrial use and power generation. Other demand centres are
the highly populated states of the northeastern US, the midwestern
states and California; Florida and the southeastern states use signifi-
cant gas generation to serve summer cooling load. Table 2.11 shows
production and estimated net exports for the main supply areas and
demand and estimated net imports for the top demand areas.
An extensive pipeline network provides for the movement of gas
from the supply to the demand areas. Many pipelines have been
built from the traditional Gulf Coast and mid-continent supply areas.
Multiple pipelines, including Texas Eastern Transmission Company
(TETCO), Transcontinental (Transco) and Tennessee Gas Pipeline
Company were built to transport gas from the Gulf states to demand
areas in the northeast. Some of these pipes are now backhauling gas
from the shale fields of the northeast back towards the Gulf. Florida
Gas Transmission and Sonat carry gas to Florida. Northern Natural
Gas, Panhandle Eastern Pipeline Company, ANR and Natural Gas
Pipeline Company of America (NGPL) deliver gas to the midwest. El
Paso Natural Gas and Transwestern Pipeline take gas west to the
California market.
The Kern River pipeline to California and the more recently built
Rockies Express Pipeline, which can move gas east to Ohio, provide
two primary outlets for gas produced in the Northern Rockies, while
Transwestern Pipeline can take San Juan Basin gas from Northern
New Mexico and southern Colorado to Arizona and California.
47
Figure 2.13 NG price (average of front 12 months) and storage levels relative to five-year trailing average (right axis)
Production declining Production growing

US$14 2,000
Recession +
mild summer
Katrina & Rita
US$12 1,600
Cold Jan-Mar
Coal to gas
US$10 switching 1,200
Mild winter '11-'12

US$/mmbtu
US$8 800
Bcf
US$6 400
US$4 -
Mild Jan
US$2 (400)
Mild summer '03 12 month strip price
Storage Delta to 5yr avg
Cold winter '02-'03
US$0 (800)
Jan-03
Jan-05
Jan-08
Apr-03
Jul-03
Apr-05
Jul-05
Jan-06
Apr-06
Jul-06
Apr-08
Jul-08
Jan-09
Apr-09
Jul-09
Jan-10
Jan-13
Apr-10
Jul-10
Jan-11
Apr-11
Jul-11
Apr-13
Oct-03
Oct-05
Oct-06
Oct-08
Oct-09
Oct-10
Oct-11
Apr-02
Apr-04
Apr-07
Apr-12
Jan-02
Jan-04
Jan-07
Jan-12
Jul-02
Jul-04
Jul-07
Jul-12
Oct-02
Oct-04
Oct-07
Oct-12
Source: EIA for reported storage
48
PRICE DYNAMICS OF GAS FUTURES

Figure 2.13 presents historical natural gas prices since 2002, and
relates how fundamental drivers of supply and demand have trans-
lated into changes of price regime. The figure shows the average
price of the front 12 futures months in order to remove seasonality
from the prices. We have reviewed earlier the most important factors
influencing the supply/demand balance which, in turn, creates pres-
sure on natural gas prices.
Let us look at some of the fundamental drivers of price levels. On
the demand side, there is:
weather (winter heating, summer cooling loads);

al price competition; and
49
US$/mmbtu
0
5
-5
10
15
20
Jan-02
Jul-02
Jan-03
Jul-03
Jan-04
Jul-04
Jan-05
Jul-05
Jan-06
Figure 2.14 Prompt month price and front-year contango
Jul-06
Jan-07
Jul-07
Jan-08
Jul-08
Jan-09
Jul-09
Jan-10
Jul-10
Jan-11
Jul-11
Prompt Futures
1 Year Contango
Jan-12
Jul-12
Jan-13
50
51
US$/mmbtu
2
4
12
0
6
8
10
Mar-02
Jul-02
Nov-02
Mar-03
Jul-03
Nov-03
Mar-04
Jul-04
Nov-04
Mar-05
Jul-05
Nov-05
Mar-06
Jul-06
Nov-06
Mar-07
Jul-07
Nov-07
Mar-08
Jul-08
Nov-08
Mar-09
Jul-09
Nov-09
Mar-10
Figure 2.15 The annual evolution of the natural gas futures curve
Jul-10
Nov-10
Mar-11
Jul-11
Nov-11
Mar-12
Jul-12
Nov-12
Mar-13
Jul-13
Nov-13
Mar-14
Jul-14
Nov-14
Mar-15
Jul-15
Nov-15
Mar-16
Jul-16
Nov-16
Mar-17
Jul-17
Nov-17
Mar-18
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002

influencing weather events are noted on the figure. A cold 2002/03

winter pushed gas storage down to very low levels and prices up
above US$6.00, before a mild summer in 2003 helped storage levels
recover, and gas sold back down to below US$5.00. Similarly, an
extremely cold January in 2008 started gas on its run towards prices
well over US$10.00. The recession of 2008 destroyed industrial
demand and sent gas prices back down, and this trend was exacer-
bated by mild summer weather in 2008 that further decreased gas
burn for power generation.
Some time periods, however, show gas prices trending generally
upward while storage also builds, such as March 2004December
2004. For most of 2006 and 2007, storage levels trended, on average,
lower, but prices gradually moved lower as well. The same happened
mid-2009 to end-2010. Referring back to Figures 2.8 and 2.11, we see
that production was on a downwards trend from 2001 to 2005, so
prices moved higher to drive out demand. Perhaps the storage builds
in 2004 were not taken as a sign of structural surplus but a temporary
respite from the tightening supply balances. In contrast, production
began its spectacular rebound in 2006, and the market took several
years to understand and digest the implications of the shale gas revo-
lution. Market prices were adjusting downward even during times
when the storage surplus was reverting to near historical levels.
Figure 2.14 shows historical prices for the prompt (ie, front)
natural gas contract. The figure also shows a measure of the
contango (slope) of the futures curve, calculated here as the price of
the 14th contract less the price of the second contract in other
words, the one-year contango of the futures curve starting at the
second to expire contract. Clearly, the level of curve contango has a
strong inverse relationship to the front month price level for most of
the 11 years of price history shown.
Because the contango of the price curve is quite volatile, traders are
attracted to trades sensitive to changes in the slope of the price curve.
Many trading strategies attempt to profit from changes in calendar
spreads by taking spread positions, shorting one month and going
long a different month. Because of the seasonal nature of gas use and
storage, certain calendar spreads tend to have more trading interest
and thus higher liquidity. Many of the favourite spreads involve the
key storage season months of October, March and April. The
March/April spread, sometimes referred to as the widow maker,
52
trades actively, as do the April/October and October/January

spreads. Two other favourites are the January/March and January/
April spreads, which have high sensitivity to winter price seasonality.
In addition to these seasonal spread favourites, the prompt/
prompt+1 month spread is active, as is the second/third futures
month spread, as index fund managers and other market participants
are active in rolling forward their nearby month positions.
Figure 2.15 shows the evolution of the front 60 months of the
natural gas futures curve. Historical curves for each year, 20022013,
are shown as of late March of each year, when the April contract is
prompt. A number of interesting features can be seen from this
evolution. The front of the curve tends to lead in most price move-
ments. Therefore, the curve will often go into backwardation when
prices move sharply higher, and contango steepens when prices
move rapidly lower. The winter to summer month spreads clearly
went higher during the high gas price environment of 20052008, but
collapsed to very low levels in 2012 and 2013.
CONCLUSION: KEY ISSUES FOR THE COMING DECADE

Since the early 2000s, the natural gas market has moved from a
period of declining production and use into a new period of produc-
tion growth so rapid that it managed to push prices back below
US$3.00, a price level that few in 2006 or 2007 ever expected to see
again. These lower prices have encouraged drillers to concentrate
more on crude oil production and less on dry gas, and at the same
time engendered a renaissance of gas-intensive industrial demand.
Increases in gas demand for industrial use and power generation
should require additional gas production, and potential exports of
LNG will accelerate demand from around 2016. At what point in
time will growth in associated gas production fail to keep up with
demand growth, requiring prices to rise to a level that will encourage
more drilling directed towards dry gas? How long will drilling effi-
ciency gains continue to push down production costs, and will
production costs begin to rise dramatically when the best shale
prospects have been produced? Even with renewed gas-directed
drilling, will production growth manage to keep up with the large
price-induced demand growth that we are witnessing?
All of these interesting questions will require constant reevalua-
tion over the coming years.
53
GLOBAL LNG
Rita D'Ecclesia
Sapienza University of Rome
Global LNG flows reached over 200 MMT in 2012, the equivalent of
almost 10 Tcf of gas or about 8% of world gas production. This panel will
discuss major exporters and importers and possible trends going forward.
Exports
By 2012, LNG exports represented about 30% of international gas flows.
Global LNG exports grew from 117 MMT in 2005 to 203 MMT by 2012,
an average annual increase of 10%. Table 2.3 shows the LNG exports for
the 10 largest exporters since 2005 including Canada scheduled to be a
major player by 2020.
The biggest LNG exporters in 2005 were Indonesia (17%), Malaysia
(15%), Algeria (14%) and Qatar (14%), accounting for 60% of world
exports. By 2012, the balance had shifted and four countries Qatar
(39%), Malaysia (13%), Australia (11%) and Indonesia (10%) accounted
for 73% of the total exports, with Algeria having heavily reduced its share.
During this period LNG exports grew by 56 MMT. In terms of
geographic distribution the Middle East was the fastest growing exporter,
growing from 38 MMT (28% of total) in 2005 to 85 MMT (43% of total) in
2012, while the Atlantic Basin reduced its exports from 44 MMT in 2005
to 37 MMT in 2012.
Exports are tied to the liquefaction capacity of each country, therefore
we need to look at the existing plants and those planned for the next
decade. In Table 2.4, the evolution of liquefaction capacity between 2000
and 2012, and an estimate for 2020, is provided. The list of exporters with
more than 10 million metric tonne per annum (MMTPA) of liquefaction
capacity is short and rapidly changing. There are 20 countries exporting
LNG and five major re-exporters (Belgium, Brazil, Mexico, Spain and the
US). Liquefaction capacity utilisation around the world averages 90%, and
so its growth is critical to expanding volumes, whereas global utilisation of
regasification is only 35%.
In 2001, the US was expected to become a major importer of LNG, but
by 2012 a resurgence in US gas production lead to the prospect of the US
becoming a major exporter once liquefaction trains become operational,
expected to begin around 2016.
Because of the high infrastructure costs of creating and delivering LNG,
most projects require long-term contracts that lock in the destination of
LNG produced. An estimated 25% of these flows are now short-term
contracts (less than four years in duration), and an increasing amount of
LNG flows are in the hands of international oil and gas companies (IOCs
see Table 2.4) with more destination flexibility.
From 2008 to 2012, IOCs increased their share of export capacity by 45
54
Table 2.3 10 largest exporters of LNG 20052015 (MMT)
2005 2008 2009 2010 2011 2012 D(20122005) 2015 D(20152012)* 2020 D(20202015)*
Algeria 15.9 15.9 15.7 14.3 12.5 11.2 4.7 19.3 8.1 19.3 0.0
Egypt 4.3 10.6 10.2 7.1 6.3 4.7 0.5 4.9 0.1 4.9 0.0
Nigeria 8.0 16.7 11.6 17.9 18.9 19.6 11.6 14.2 5.4 14.2 0.0
Oman 5.7 8.6 8.1 8.6 8.1 8.2 2.4 8.3 0.2 8.3 0.0
Qatar 16.8 30.0 36.9 56.2 75.4 76.4 59.6 75.3 1.1 75.3 0.0
Australia 9.2 15.0 17.9 18.8 19.5 20.9 11.7 21.7 0.8 77.3 55.6
USA 1.1 0.8 0.6 0.6 0.3 0.2 1.0 9.9 9.7 80.8 70.9
Indonesia 19.5 20.1 19.3 23.5 21.9 19.0 0.5 13.6 5.4 15.1 1.5
Malaysia 17.6 22.1 22.3 23.2 24.9 24.9 7.3 25.9 1.0 25.9 0.0
Russia 5.0 9.9 10.6 10.9 10.9 9.6 1.3 9.6 0.0
Canada 16.9 0
World total 117.0 139.8 147.5 180.0 173.5 195.9 56.0 202.6 22.6 347.5 144.9

Major exporters D(20122005) D(20152012)* D(20202015)*
Maj Pac Basin 57.2 64.5 75.3 76.9 75.6 18.4 70.7 4.6 144.8 74.0
% of total 41% 44% 42% 44% 39% 35% 42%
Middle East 38.6 45.0 64.8 83.5 84.5 45.9 83.7 18.9 83.7
% of total 28% 31% 36% 48% 43% 41% 24%
Maj Atl Basin 44.0 38.0 39.9 38.0 35.7 8.3 48 8 119 70.9
% of total 31% 26% 22% 22% 18% 24% 34%
* Estimates by GIIGNL.
55
billion cubic meters per annum (bcm/a) from 85 to 130 bcm/a, led by
Shell, Exxon Mobil, Total, ConocoPhillips, Woodside and Chevron.
National oil and gas companies (NOCs) increased by 74 bcm/a, from 137
to 211 bcm/a. Trading houses, LNG importers, financial institutions and
local companies represent the balance, 33 bcm/a in 2008 and 48 bcm/a
by 2012. NOCs have an obligation to satisfy domestic demand, therefore
Russia, Nigeria and Indonesia are increasingly focused on the price gap
between their domestic market and export prices. In general, IOCs are
more responsive to market conditions, and bring advantages in terms of
integrated project development. European utilities with considerable LNG
strategies include GDF-Suez, EdF, E.ON and RWE.
In 2012, Qatar dominated global export capacity with a 39% market
share and 84 MMTPA of liquefaction (see Table 2.5). The other Middle
East exporters, including Abu Dhabi, Oman and Yemen, have no reported
plans to expand their liquefaction capacities. Qatar is a true swing
exporter and, in the period 200812, sent on average 35% to Europe, 5%
to the Americas and the rest to Asia (of which 33% was to Japan, 25% to
each of India and South Korea, 10% to Taiwan and 7% to China). Asian
demand growth is impressive (see Table 2.6). China has grown from
nothing in 2005 to 5 MMT in 2012, India from 6 to 10 MMT, Japan from 8
to 16 MMT and Taiwan from 1 to 6 MMT. South Korea is the only stagnant
Asian importer, with 9 MMT in 2005 and 11 MMT in 2012. Most of the
LNG from Abu Dhabi, Oman and Yemen flows to Asia.
The Pacific Basin liquefaction capacity stands at 92 MMTPA, repre-
senting 38% of the world total. It is expected to increase by 2020 as many
large Australian and Canadian projects come online, and Australia is
expected to tie with Qatars liquefaction capacity. Indonesia has been
experiencing domestic production outages, and is therefore planning to
expand its liquefaction capacity to send out 40% of production to the
domestic market. In addition, adding new liquefaction capacity in 2014,
Indonesia is converting two ageing liquefaction plants to regasification.
Malaysia has had a series of outages on liquefaction maintenance and has
minor plans for floating liquefaction in the future.
Australia and Canada are positioned to be key exporters in this basin. In
the period 200512, Australia added 24 of the 26 MMTPA Pacific Basin
liquefaction increase. According to planned new liquefaction plants,
Australia will increase its capacity by 60 MMTPA, and Canada is expected
to build 17 MMTPA of liquefaction capacity by 2020, estimated as 50% of
the 34 MMT of filed projects.
The major Atlantic Basin exporters hold 23% of liquefaction capacity.
From 2005, Algerian capacity has remained unchanged at 19 MMTPA,
still recovering from the 2004 explosion at Skikda that kept capacity
offline in the 200812 period. New capacity additions for Algeria have
been quoted at US$1,000/MT capital costs. Egypt started as an exporter in
2004 and has 12 MMTPA of capacity. Its economic growth has created
more domestic demand, and it is planning to build regas capacity. By
56
Table 2.4 Liquefaction capacity (MMTPA) (estimates by the author).
Country Basin 2000 2005 2008 2009 2010 2011 2012 2015 2020
Algeria Atlantic 19.4 19.4 19.4 19.4 19.4 19.4 19.4 8% 24.1 9% 24.1 6%
Egypt Atlantic 0 12.2 12.2 12.2 12.2 12.2 12.2 5% 12.2 5% 12.2 3%
Nigeria Atlantic 9.6 9.6 21.8 21.8 21.8 21.8 21.8 9% 21.8 8% 21.8 5%
Oman Middle East 7.1 7.1 10.7 10.7 10.7 10.7 10.7 4% 10.7 4% 10.7 3%
Qatar Middle East 16.1 25.5 36.9 60.3 75.9 83.7 83.7 35% 83.7 33% 83.7 20%
Australia 1 Pacific 0 12.1 19.8 19.8 19.8 19.8 24.1 10% 24.1 9% 85.9 21%
USA2 Atlantic 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1% 10.4 4% 85 21%
Indonesia Pacific 26.5 26.5 26.5 34.1 34.1 34.1 34.1 14% 33.95 13% 37.75 9%
Malaysis Pacific 15.9 22.7 22.7 22.7 24.2 24.2 24.2 10% 26.07 10% 26.07 6%
Russia Pacific 9.55 9.55 9.55 9.55 4% 9.55 4% 9.55 2%
Canada3 Pacific 16.9 4%
100% 100% 100%
TOTAL 96.0 136.5 171.4 212.0 229.1 236.9 241.2 256.6 413.7
200500 200805 200908 201009 201110 201211 201512 202015

Capacity change 40.5 34.9 40.55 17.1 7.8 4.3 15.42 157.1
and percentage
share
Capacity by area 2000 2005 2008 2009 2010 2011 2012 20122005 2015* 20152010 2020* 20202015
Maj Pac Bas 42.4 61.3 69 86.15 87.65 87.65 91.95 26.35 93.67 1.72 176.17 82.5
% of total capacity 44% 45% 40% 41% 38% 37% 38% 37% 43%
Middle East 23.2 32.6 47.6 71 86.6 94.4 94.4 54 94.4 0 94.4 0
Maj Alt Bas 30.4 42.6 54.8 54.8 54.8 54.8 54.8 12.2 68.5 13.7 143.1 74.6
1
Adds 10 MMTPA capacity, or more.
2
134.2 Mmtpa fled with FERC.
57
3
Canada is expected to build in 2015, the 50% of 34Mmtpa in liquefaction capacity (estimates by the author).
many estimates, Egypt will not be an exporter by 2020. Its utilisation of

liquefaction has dropped from almost 90% in 2008 to 40% in 2012 (see
Table 2.11). Nigeria holds 22 MMTPA of liquefaction, more than doubling
since 2000, but suffers considerably from political unrest and infrastruc-
ture construction delays. Despite having the greatest gas capacity in the
Atlantic Basin, it continues to struggle to perform. The US is expected to
operate 85 MMT of liquefaction capacity by 2020 out of the 135 MMT of
filed projects, according to the authors estimates.
Importers and import growth

Import demand is relatively simple to analyse in the LNG market, given
the different regional demand drivers. Asia depends heavily on oil, and
LNG increasingly flows to the industrial complexes on the southern coast
of China. Chinas natural gas assets are in the Northwest, and while the
trans-China gas pipelines will inevitably be built, LNG is at least the short-
Table 2.5 Regasification capacity by country, 200020 (MMT)
Country 2000 2005 2008 2009 2010 2011 2012 2015* 2020*
Belgium 4 4 4 4 4 4 4 9 10
France 7 7 7 11 11 11 11 20 22
Italy 2 2 2 5 5 5 5 16 27
Netherlands 5 5 15 20
Spain 19 22 27 27 27 27 27 27 27
Turkey 3 3 6 6 6 6 6 6 10
UK 9 11 24 24 24 24 24 27
Big 7 total 34 47 57 77 77 82 82 117 143
Europe 34 47 59 77 77 84 88 125 150
USA 3 8 46 53 78 83 83
Americas 3 8 50 60 100 110 112
China 8 10 10 12 14
India 5 5 6 8 8 8
Japan 104 108 115 115 116 117 118
South Korea 44 55 55 55 55 55 55
Taiwan 4 4 4 7 7 7 7
Asia 152 168 175 177 178 179 207
Middle East 4
Total 189 223 284 314 355 373 411
* Estimates by GIIGNL.
58
term supply choice. Coastal India is another big importer, where GDP is
crimped by a lack of energy, and rolling brown-outs are common.
Asia more than doubled imports in the period 200512, with Indonesia
and Taiwan starting to import in 2005, with Japan, China and South Korea
also increasing their volumes. Europe and the Americas reduced their
LNG imports in 201012, despite increasing between 2005 and 2010
(Table 2.6), due to factors such as price, the economic downturn and
increasing US domestic production.
Asia is the largest importing region, with almost 65% of total world
imports. In 200812, Asia imported an average of 136 MMT (63% of
world total imports). Of these, 55% was delivered to Japan, 22% to South
Korea, 6% to China and the remaining 17% to India, Taiwan and
Indonesia. Imports in Asia have staged a recovery after a contraction in
2009 (7%, see Table 2.7).
European imports increased by 70% during 200512. In 2012, they
accounted for 21% of global imports. The largest importer is Spain (31% in
2012), followed by France (15%), the UK (21%), Turkey (11%), Italy
Table 2.6 LNG imports by country (MMT)
2005 2008 2009 2010 2011 2012 D(20122005)
Belgium 2 2 5 5 5 3 1
France 8 9 10 10 11 7 1
Italy 2 1 2 7 6 5 4
Netherlands 1 1 1
Spain 14 22 20 21 17 15 2
Turkey 3 4 4 6 5 5 2
UK 0 1 7 14 19 10 10
Big 7 Total 28 40 48 62 63 47 19
Europe 30 42 52 65 65 49 20
USA 11 7 10 9 6 4 8
Americas 12 11 16 21 19 18 6
China 3 6 10 13 15 15
India 3.7 8 9 12 12 13 9
Japan 47.2 69 66 72 78 87 40
South Korea 18.8 29 21 28 35 37 18
Taiwan 5.9 9 9 11 12 13 7
Asia 75.7 118 113 132 153 166 90
Middle East 0 0 1 2 4 3 3
TOTAL 117.7 195 181 220 241 236 119
59
Table 2.7 LNG imports by country (growth rate %)
2008/2005 2009/2008 2010/2009 2011/2010 2012/2011
Big 7 Europe 40% 22% 29% 0% 25%
Europe 43% 23% 25% 0% 24%
Americas 2% 37% 33% 8% 5%
China 68% 70% 37% 13%

India 119% 13% 35% 1% 5%
Japan 45% 4% 9% 9% 12%
South Korea 53% 27% 32% 26% 4%
Taiwan 54% 3% 21% 9% 8%
Asia 56% 4% 17% 16% 9%
Total 67% 7% 21% 9% 2%
(10%), Belgium (6%) and the Netherlands (1%). These six countries
account for the lions share of demand (96%).
Imports by the Americas accounted for an average 17 MMT over 2005
12, and in 2012 were a mere 7% of world LNG imports. The US
accounted for 44% of the volume followed by Mexico (19%), Argentina
and Chile (9% each), and 6% for Brazil and Canada.
Two countries in the Middle East (Kuwait and Dubai) started to import
LNG in 2009 and in 2012 were an insignificant 1% of global imports.
Imports of LNG are linked to the regasification capacity of the various
importing countries (see Table 2.8). In 2012, there were 93 LNG regasifi-
cation terminals operating in the world including 11 floating facilities.
There are two possibilities for significant regasification capacity growth
around the world. Both China and India have considerable plans to
expand LNG imports. The GIIGNL 2012 Annual Report lists eight projects
under development in China that are expected to add some 15 MMT of
regas capacity. This would double Chinese import capacity. By 2020,
China could be importing as much as South Korea. India similarly lists
12.5 MMT of capacity under construction, likely to continue to be
hampered by logistical issues, and also lists a variety of terminal and distri-
bution projects. This would more than double Indian import capacity into
the early 2020s.
Regasification in Europe is mainly concentrated in the seven largest
European importers which have 82 of the total 88 MMTPA of regasifica-
tion capacity. The utilisation rates swing depending on the LNG price. For
example, Spain imported 22 MMT in 2008 and only 15 MMT in 2012. The
flexibility of imports in Europe is a reflection of its market maturity and effi-
ciency. The LNG demand in Europe has been growing at a fast pace over
200510, with an annual average growth of 19% to 2011, and declined
60
Table 2.8 LNG regasification capacity by country (percentage of utilisation)
2008 2009 2010 2011 2012 2015* 2020*
Belgium 56 121 112 115 73 95 95

France 132 89 96 97 68 96 96
Italy 72 42 130 122 98 93 93
Netherlands 13 10
Spain 81 75 76 62 56 70 70
Turkey 70 73 100 85 98 85 85
UK 7 30 60 79 44 44 44
Big 7 total 70 63 81 76 57 69 69
Europe 71 68 56
USA 16 18 11 7 4 11 11
Americas 16
China 41 56 95 110 107 82 82

India 154 155 157 156 164 157 157
Japan 60 57 62 67 74 64 64
South Korea 52 38 50 64 66 54 54
Taiwan 211 130 158 172 186 171 171
Asia 80 85 90
Middle East 72 75 80
Total 35 40 40
Source: Authors estimate
heavily in 2012 (25%, partly due to relative price and partly economy
shrinkage). The large reduction of LNG demand is in line with the heavy
reduction of natural gas demand in 201112 in Europe. In the period
200508, virtually every European country, from Lithuania to Ireland,
added regas capacity.
In the US during 200508, a lot of regas capacity was built, but subse-
quently was not needed, so US utilisation rates are abysmal.
Regasification global usage is only 35%, but capacity utilisation varies
widely by region. Regas capacity can provide flexibility and security of
supply. Utilisation rates change as capacity is added and as other energy
flows dictate. For example, between Spain and France energy may flow as
gas or be wheeled as electricity.
Table 2.9 lists regasification capacity utilisation rates. Italys data is diffi-
cult to follow, with listed additions apparently running earlier than official
openings. India suffers from the same problem. Russia and China do run at
excess of nameplate capacity. Taiwans import data is suspect. All figures
come from GIIGNL.
61
Table 2.9 European natural gas supply and demand in the European Union (bcm)
2000 2005 2008 2009 2010 2011 2012
Production 23193 21198 19328 17426 17779 15793 14965
average (200812) 1706
Consumption 44029 49613 49729 46512 50289 45305 44388
average (200812) 4724
Russian pipeline imports 19390 15128 18099 16429 18634 17856 18590
average (200812) 1792
Excess demand (BCM) 1447 13288 12302 12657 13876 11656 10833
average (200812) 1226
Excess demand in LNG equivalent (MmT) 1070 9833 9103 9366 10269 8626 8016
average (200812) 908
LNG supply chain

The cost of gas is critical to the analysis of future export availabilities, espe-
cially for US shale gas. The cost of building liquefaction has risen
dramatically:
the variable costs of liquefaction in the US are approximately

US$2/MMBtu;
transatlantic freight is approximately US$1/MMBtu; and
regas costs are US$0.50/MMBtu.
This means a built-in supply cost which must be added to the natural gas
price (Henry Hub) of US$3.50/MMBtu for gas landed into Europe. This
natural gas chapter estimates the price at which we will continue to
expand US shale gas at US$4.005.00/MMBtu leaving us with a landed
Europe price of US$7.508.50/MMBtu. Notwithstanding this high price,
we expect to see a continued healthy European demand, especially if GDP
growth can recover.
Assuming Japan has an incremental freight cost of US$2/MMBtu the
natural gas price for Asia may reach US$10.50/MMBtu. The liquefaction
cost in Canada, after the building of the planned liquefaction plants, is
expected to be close to US$1.70/MMBtu, and these volumes are directed
to Japan.
Terminal expansion is lowering costs along with expanding fleet and
vessel size. The global fleet is 378 vessels and 54,000,000 m3, including
floating storage and regasification units (FSRUs). Only two vessels were
added to the fleet in 2012, compared to 16 in 2011, three ships were
scrapped and one was converted to an FSRU. More than 40 vessels in the
fleet have been used for over 30 years and more than 250 vessels are
under 10 years old.
62
The order book was 78 vessels at the end of 2012 and 27 new orders
were added in the year, of which 23 were LNG carriers ranging from
150,000172,000 m3, two FSRUs, one regasification vessel (RV) and one
floating liquefied natural gas (FLNG) carrier (210,000 m3). ICIS Heren has
forecasted that additional expansion is needed for the fleet in order to
retire older ships in 201520.
What may be more important for estimating future shipping flows is the
ever-increasing share of flows to the Pacific basin, rather than the Atlantic
basin, lengthening tonne miles. The future growth of European demand,
on the other hand, depends mostly on building storage and distribution
assets, where environmental and other compliance issues will be consid-
erably more expensive than in emerging or frontier markets. Concerns
over emissions seem to be curtailing European demand for LNG and
compressed natural gas (CNG) as a truck fuel.
Future LNG flow considerations

Liquefaction plant build costs in the early 2000s (such as Egypts US$250
350/MMTPA and Omans US$200/MMTPA) were comparatively low.
Qatar RasGas II and III build costs were around US$350/MMTPA, while
Qatargas IV was close to US$750. Australian Pluto was estimated at
US$800 and the Russian Sakhalin capacity got deferred on an estimated
US$1,000. Geography, climate and political risks drive construction
costs. An ever-increasing amount of gas trying to come to market from
emerging countries (Equatorial Guinea, Yemen, Peru, Angola, PNG, Libya
and Iran) will not help lower costs of future liquefaction capacity addition.
This will make it increasingly easier for an IOC to get involved, compared
to an NOC.
More generally, domestic gas demand is growing in many producing
countries for generating power and water, fuels and petrochemical
production, as well as reinjection to oilfields.
In terms of major exporters, we note that Qatar, who have paused lique-
faction at current levels, actually have approvals in place to expand
liquefaction up to 105 MMT. This represents an opportunity. Nigeria still
has considerable waste between gas field and liquefaction, and an uncer-
tain future for further developing gas pipelines within Africa. The US has a
major opportunity to capture export market share, but energy exports have
no great historical precedent within the worlds largest energy consumer.
Russia will inevitably add more LNG capacity for Asia.
Europe has experienced a reduction in natural gas production since
2000, from 232 bcm in 2000 to 150 bcm in 2012 (see Table 2.9). The
demand for natural gas reached a high of over 500 bcm in 2010, but by
2012 was back near the 2000 level of 440 bcm. Russian natural gas
pipeline exports to Europe have declined since 2000 bringing an increase
in other import demand from 14.5 bcm in 2000 to 108 bcm by 2012. This
equates to a European LNG demand of 80 MMT in 2012.
63
Table 2.10 Total LNG capacity holders (bcm/year)
2008 2012 D(20122008) D%
IOCs
Shell 19.3 27.4 8.1 42
BP 15.3 17.3 2.0 13
BG 9.7 9.7 0.0 0
ExxonMobil 9.3 20.8 11.5 124
Total 7.9 14.6 6.7 85
ENI 6.3 7.3 1.0 16
Repsol/Gas Natural 4.7 5.9 1.2 26
ConocoPhillips 4.0 7.2 3.2 80
Marathon 3.4 3.4 0.0 0
Woodside 2.7 9.6 6.9 256
Chevron 2.7 6.3 3.6 133
TOTAL 85.3 129.5 44.2 52
NOCs
Pertamina (Indonesia) 39.6 39.6 0.0 0
Qatar Petroleum 27.8 84.0 56.2 202
Sonatrach (Algeria) 27.8 33.9 6.1 22
Petronas (Malaysia) 25.4 26.5 1.1 4
NNPC (Nigeria) 14.8 14.8 0.0 0
StatoilHydro (Norway) 1.9 1.9 0.0 0
Gazprom 0.0 10.0 10.0 10+
TOTAL 137.3 210.7 73.4 53
Table 2.11 Percentage plant utilisation
2008 2009 2010 2011 2012 2015 2020
Algeria 82% 81% 74% 64% 58% 80% 80%

Egypt 87% 83% 58% 52% 39% 40% 40%
Nigeria 77% 53% 82% 87% 90% 65% 65%
Oman 81% 76% 81% 76% 76% 78% 78%
Qatar 81% 61% 74% 90% 91% 90% 90%
Australia 76% 90% 95% 98% 87% 90% 90%
USA 56% 43% 44% 21% 12% 95% 95%
Indonesia 76% 57% 69% 64% 56% 40% 40%
Malaysia 97% 98% 96% 103% 103% 99% 99%
Russia 0% 53% 103% 111% 114% 100% 100%
Canada 100% 100%
Total 100% 100%
Source: Authors estimate
64
3
A Day in the Life of Commodity
Weather
Jose Marquez
Whiteside Energy
This chapter will offer insight into the role of a commodity meteorol-
ogist and how they aid our understanding of risk within commodity
markets. Primary sources of information, methods of interpretation
and strategy considerations are given from the perspective of an
energy trading firm. Weather linkages in other commodity markets
are also briefly discussed.
Weather drives daily volatility demand for natural gas. Weather
influences residential, commercial and electrical power end users,
natural gas is burned in the winter for heating and electrical genera-
tion requirements in summer. Regional demand differences and
seasonality ultimately affects natural gas futures pricing and
regional basis hubs. In natural gas markets, cold weather can force
peak day demand events where price-induced curtailments may
occur to non-temperature sensitive clients (ie, reduction of industrial
load) in order to ensure that needed gas is available to residential and
commercial consumers. Residential and commercial sectors require-
ments peak during the heating season, and gas must be stored to
meet the winter demand.
Weather is a constant source of short term volatility in natural gas
demand and price expectations. Therefore, a solid understanding of
the relationship between weather and natural-gas fundamentals is
imperative.
65
WEATHER DATA BASICS

Meteorologists working for commodity trading firms have long been
utilised in agriculture markets, where extreme weather conditions
affect diverse crops throughout the year. The US National Weather
Service (NWS) and several weather consulting firms provided
weather information and forecasts for 15 and 610 day periods.
Meteorologists then enhanced this information through further
interpretation and acted as a quality control for the weather forecasts
provided by these external sources. In the early 1990s with the dereg-
ulation of natural gas, Enron was the first energy merchant to utilise
meteorologists on staff to expedite and maximise the accuracy of
weather forecasts. The company understood the significant correla-
tion between temperature and natural gas demand, and that being
ahead of the pack at incorporating incoming temperature changes
would help maximise profits on their large natural gas portfolio. For
example, buying or selling natural gas molecules ahead of others
gave the ability to profit from expected increase or decrease in
demand, which then moves price on a regional or national basis. Of
course, such methods to create a trading edge do not last forever.
Soon, many other energy trading firms maintained their own staffs
of in-house meteorologists. At one point, Enron had a team of six
people providing weather information to the trading desks.
The main daily source of weather information for everyone across
the globe comes from global weather models. Some models provide
forecasts up to 10 days, others up to 16 days. In a nutshell, a global
weather model is a sophisticated mathematical model that uses a set
of equations with diverse parameterisations that represent the Earth
and atmosphere. Horizontally and vertically, the Earths surface and
the atmosphere is divided into grids or pixels that interact with each
neighbouring point, ultimately allowing calculation of a forecast for
the future state of the atmosphere. The resulting forecast may step
through time, starting with three-hour increments increasing to 12-
hour time steps after 192 hours. The size of the geographic and
temporal grids are tuned in order to optimise the balance between
the number of required computations and the grid resolution, since
more calculations are required using the higher resolution grid
compared to the lower resolution grid. As an example, the grid size
or pixel may vary from 35-kilometre spacing up to 70-kilometre
spacing for forecast periods after 192 hours (8 days).
66
A DAY IN THE LIFE OF COMMODITY WEATHER
Meteorological and oceanographic data to initialise the models

come from across the globe: from air and land weather recording
stations, weather satellites and commercial and military pilot
reports. This immense dataset is gathered, assimilated and fed into
various global weather models. An initial condition or initialisation
defines the beginning state of the earthatmosphere system, before
forecasts with a defined time stamp are calculated by the models. As
you can imagine, the amount of data and the computational power
required to run these models are immense, and to truly obtain an
adequate global initial condition requires full access to global data
(some data could be considered confidential). Consequently,
specialised government agencies or research centres with special
international agreements for data sharing are the only entities
capable of producing a meaningful and skillful global forecast.
Therefore, meteorologists across the world obtain their daily
temperature and weather changes from global weather models
produced by various institutions. In the energy industry, the main
models observed and analysed are the American Model (GFS), the
European Model (ECMWF) and the Canadian Model (GEM). In
addition, and to a lesser extent, there is the NOGAPS (US Navy) and
short-range models such as the NAM (up to 84-hour forecasts). The
American model is run by the US National Weather Service's
National Center for Environmental Prediction, in Washington DC,
the European model is run by the European Centre for Medium-
Range Weather Forecasts, located at Reading, UK, and the Canadian
Model is run by Environment Canada (Canadas National Weather
Service).
The US National Weather Service provides daily forecasts for the 1
5, 610 and 814 day periods. The information comes in data output or
graphical format.
A DAY IN THE LIFE

Early in the morning, multiple weather sources release information
which could be utilised by the markets. The changes on weather
information and data are compared to the previous trading day
determining upcoming changes in natural gas demand and setting
the tone for traders early in the morning. Traders know that colder
than normal conditions in the highest population areas of the US,
mainly East of the Rockies in the winter means higher demand
67
for the US as a whole. In summer months, warmer than normal

temperatures in the East, especially Texas and Southeastern US,
means more demand for air conditioning, of which a large
percentage is generated by natural gas-fired power plants.
Meteorologists on staff do not influence the market with their
information or have an influence on Nymex pricing. Their informa-
tion is kept in-house. On the other hand, weather information and
forecasts come from multiple sources, including global weather
models which have a broad dissemination across markets. Thus,
large changes to the forecasts can create a tangible reaction in the
energy markets.
Meteorologists have their own language to forecast or explain
weather patterns and/or phenomena. They talk in terms of geopo-
tential heights, vorticity and jet streams to mention a few. Energy
traders talk in terms of Heating/Cooling Degree Days (HDDs/
CDDs), increase/decrease demand, confidence level and risks.
Therefore, the most important job of the in-house meteorologist is to
"translate" the meteorology language into an energy trader's
language. They link the language of science to trading. The meteo-
rologist on staff will gather all relevant information available from
multiple sources and streamline it in a way that is easily accessible
and understood by the trading desks. The meteorologist could come
with the following checklist: How is the weather pattern evolving for
the 610 and 1115 day periods? What is my confidence level in the
weather pattern? What is the risk of the forecasts to change direction-
ally and temporally? The in-house meteorologist gives a sense of
confidence level for the existing forecast. If the staff meteorologist
feels that the current forecast may change then forecasting how that
change is likely to occur, in timing and direction, becomes critical.
First, the meteorologist on staff has their own view of the weather
pattern for the 610 and 1115 day periods. When all the moving
parts are in agreement, for example when diverse global weather
models forecasts are aligned, the job for the in-house meteorologist is
usually uneventful. However, when the in-house meteorologist is in
disagreement with the diverse global models output, the situation
can be quite challenging.
Most of the time, the divergence in forecasts starts when global
weather models are differing in their output. For example, the
European model may be showing a cold wave in the Midwest while
68
the American model does not show it for the same time period. So,
there is no middle ground here and a forecast must be made. Does
the Midwest have a cold event or not? Therefore, the in-house
meteorologist has to react with a highly accurate, timely response
and be prepared to accommodate many information requests from
traders.
An important process after having a forecast view of the incoming
weather pattern is anticipating how or when the forecasts from
various sources may change. This task is called "forecast the fore-
cast". Overall, agreement or disagreement with the forecast's output
from various sources serves as a confidence level barometer for
traders. Situations arise when the Nymex price moves strongly due
to forecasts of impending cold or warm events, and traders can put
immense pressure onto the in-house meteorologist to either change
the internal forecast or to precisely time when the forecasts will
change. Therefore, it is the meteorologists job to make such infor-
mation both accessible and easy to understand, and to be clear and
concise about the risks from a challenging forecast.
Following Keynes advice that Wordly wisdom teaches that it is
better for the reputation to fail conventionally than to succeed uncon-
ventionally, the easiest way out is to agree with the general weather
view of the markets, and when the pattern surprisingly changes,
then point to the fact that global weather models were wrong. To
provide true value to the firm, however, the meteorologist must
make the best possible assessment of forecasting the forecast revision
and communicate that opinion along with the relevant risks to the
trading desks. The meteorologist should not get overly bogged
down in details but focus on the importance of getting the weather
pattern right first, and then worry about the details. Simpler is better.
After the early morning weather operations are finalised, several
weather updates will arrive during regular trading hours. As the
numerical models update, any significant change in the weather
pattern compared to early morning weather information could cause
price volatility. The NAM is the first one to update, although this
weather model only provides forecasts up to 3.5 days ahead. The
GFS is the first global weather model to update the 16-day forecasts.
The GFS is immediately followed by its ensembles, a package of fore-
casts that show the level of stability or instability of the current
solution. Then, the ECMWF updates after the American models are
69
done. The whole updating process of new weather information

consumes the last three hours of the regular Nymex trading day.
TROPICAL WEATHER
There is a seasonal weather system that creates quite volatile price
action during the summer months: hurricanes. The hurricane season
runs from June 1 until November 30 in the Atlantic Basin. The main
threat area is the Gulf of Mexico, specifically from Mobile to just
north of Corpus Christi. Historically, close to 10% of total gas
production in the US could be impacted. The National Hurricane
Center (NHC), is the official entity responsible for issuing tropical
forecasts, watches and warnings.
NHC establishes a tropical cyclone as an organized system of
clouds and thunderstorms with a low level circulation rotating anti-
clockwise in the Northern Hemisphere. Tropical cyclones develop
over tropical or subtropical waters. They are classified as follows:
Tropical Depression: Maximum sustained winds of 33 knots or

less;
Tropical Storm: Maximum sustained winds between 34 to 63
knots. At this level, tropical cyclones are named; and
Hurricane: Maximum sustained winds greater than 64 knots.
A hurricanes exact centre location can easily be identifiable via

satellite imagery because of the development of an eye. In addition, a
hurricane wind scale called the SaffirSimpson is used to classify
hurricanes into five categories depending on their wind intensity.
Category 1 hurricanes are dangerous and create some damage, while
category 5 hurricanes are monster storms that create catastrophic
damage. A storm is classified as a major hurricane when it reaches
category 3 or higher. In terms of the energy markets, the biggest
concern is when the hurricane becomes a major hurricane. At this
level, structural damage to energy infrastructure may occur both
offshore and onshore. Rigs and platforms can be destroyed and
severe damage may be inflicted on onshore refineries. Underwater
pipelines can also sustain damage due to heavy wave activity.
The NHC naturally has human safety as its primary objective, and
has been designated at the one official source of forecasts in order to
reduce possible confusion during hurricane events. History has
70
shown that conflicting forecasts and "hype" from different media

outlets creates public confusion as well as potentially causing confu-
sion in the energy markets. Imagine if there were several scientific
and media venues with different forecasts and weather/hurricane
model solutions showing landfall of specific hurricane ranging from
North Carolina to Tampico, Mexico. NHC is the liaison for all the
data gathering, scientific streamlining, government safety planning
and coordination, and dissemination of information to keep the
public alert and informed. When a tropical cyclone develops, they
send standard advisories every six hours, at 0300, 0900, 1500 and
2100 UTC, that include up to five days of forecast information. When
the tropical cyclone reaches a level of tropical storm or hurricane and
may be impacting land in the next 48 hours, watches and warnings
may begin to be issued, and intermediate advisories are released
every three hours between the main advisories after a watch and/or
warning has been issued.
Imagine such a large system being modelled mathematically,
trying to represent the entire structure and energy of the tropical
system. That is what global weather and hurricane models try to do.
As would be expected due to limited numerical capacity and
inherent model limitations, different models will show somewhat
different forecasts and, even worse, may show quite different fore-
cast tracks for the storm. Global models may start by showing a
tropical system developing on day 16 off of the West Coast of Africa,
and Nymex price action may start to be influenced by the forecast. In
this scenario, three basic questions should be asked: Is the tropical
system going to develop into a hurricane? Will it be a threat to the
Gulf of Mexico? Most importantly, is it likely to grow into a major
hurricane that can damage infrastructure?
Therefore, from the NHC advisories and the constant flow of
updated hurricane output solutions from the models, the markets
become quite jittery, reacting to the diverse information as it is
revealed. If all models show the hurricane moving to the open waters
of the North Atlantic, the market will see that as a 0% chance of
impacting production. However, if one of the global or hurricane
models shows the hurricane moving into the Gulf, there is a chance
of a market-moving event which will be reflected in the price
action.
The in-house met has to constantly monitor all the information,
71
analyse all the forecasts available and, of course, forecast the forecast
of the official tropical NHC advisory. The time between a tropical
cyclone developing off the coast of Africa and reaching the Gulf of
Mexico can take nearly ten days. High volatility of energy prices
comes packaged with these systems and persists over the lifetime of
these tropical cyclones.
OTHER WEATHER IMPACTS

Weather updates during regular trading hours provide energy
traders with significant demand change expectations for North
America down to a regional and individual city level. In the summer,
power traders are the most sensitive to small changes in tempera-
ture, cloud cover, precipitation and wind. Sea breezes or
thunderstorms over downtown cities create rapid and significant
changes in electricity demand. Therefore, meteorologists providing
information to power traders have to be in tune with radar and satel-
lite images on a constant basis during the trading day.
Agriculture
Reuters, May 2013: After a cold and wet spring in most of the US
crop belt, farmers have seeded 28% of their intended corn acres, up
from 12% a week earlier but far behind the five-year average of 65%,
Chicago Board of Trade corn and soybean futures were trading
higher on Tuesday, due in part to the slow planting pace that threat-
ened to trim 2013 production prospects.
October 9, 2012, the Financial Times reported that hopes for bountiful
crops in South America fell after forecasts reduced the likelihood of
El Nio conditions developing, reducing the probability of above-
normal rains during the growing season.
Bloomberg reported on May 2, 2013, Oklahoma wheat production,

already expected to decline 45% from a year earlier, may fall further
as freezing weather tonight threatens crops.
September 12, 2012, The New York Times story, US Lowers Forecast
of Crop Yields for a 3rd Time as Record Heat Lingers, reported that
the USDA lowered forecast corn and soybean yields as record heat
added to drought damage.
The normal daily meteorological operations used in the energy busi-

ness can be extrapolated to other commodity markets for which
weather changes/influences the supply or demand for a commodity.
The most obvious are the agriculture markets. The planting season
72
for corn or soybeans could be delayed or run ahead of time

depending on spring temperatures and rainfall. Too much rainfall
does not allow planting processes to take place on muddy fields. In
addition, corn needs a minimum of 50F and adequate moisture for
germination. If soil temperatures remain below 50F after planting,
damage to the corn seed can be severe. Therefore, the germination
process could be curtailed. A cold spring, such as the spring of 2013,
will delay the planting season and make the corn more susceptible to
summer heat during pollination. In the summer, drought conditions
and temperatures above 95F with low humidity can cause damage
to the exposed silks, potentially damaging pollen. During this
period, weather forecasts of potential heatwave across the US Corn
Belt can create a quite volatile price action in the corn market.
Transport
On January 4, 2013, Time reported that drought conditions could
disrupt barge traffic on the Mississippi river, disrupting corn,
soybean and grain transport.
Drought conditions in the Midwest and Ohio Valley can affect the
river levels at the Mississippi and Ohio rivers. Coal and agricultural
barges might be restricted from travelling across the low levels of
these rivers. Supply of coal and agricultural goods could be affected
on a regional basis due to transportation restrictions. Even nuclear
power plants can be affected by drought conditions: nuclear facilities
need large amounts of water for cooling purposes. After the water
has been utilised in the plant, it is discharged back to a nearby body
of water at a higher temperature. State and federal regulations
prohibit nuclear plants from continuing operations once the water
temperature reaches a certain threshold. There is a two-fold issue
here: it compromises the reactor safety and affects aquatic life.
Livestock
January 2013, Bloomberg reported, Hogs futures climb as US cold
may hinder supply, noting that Northern temperatures of 1015F
might disrupt the movement of animals to market.
May 2, 2013, Farmers Weekly reported that UK livestock deaths
exceeded 100,000 because of March blizzards and extreme freezing
weather.
73
A cold wave creates stress in cattle, despite the bovine being

extremely tolerant to low temperatures. An adequate winter coat
and body condition in addition to availability of food and water help
them to withstand the cold. However, the bovine will lose body fat
during a cold event and in many severe cold temperature events,
hypothermia and death can occur. Newborn calves are also at high
risk of death during cold weather events. The cattle markets typically
react in quite a volatile way when these weather events occur in the
Texas/Oklahoma Panhandle and lee side of the Rocky Mountains.
Softs
May 29, 1997, The New York Times reported that Fears of Freeze in
Brazil Push Coffee Prices to 20-Year High.
July 3, 2009, Bloomberg reported that cocoa crops in Indonesia and
Ecuador could be damaged by El Nio conditions, bringing lower
rainfalls.
Coffee futures can become quite volatile if strong cold events affect
Southern Brazil. Brazil is the largest coffee producer and the only one
threatened by frosts. The coffee plant cannot tolerate frost. Depending
on frost intensity, the flowers get killed or the entire tree can die. If the
plant dies, then new plants need to be planted and it can take around
three years for them to bear coffee cherries. Vietnam is another large
producer of coffee but the main weather threat to coffee production is
the landfall of typhoons into that country. Cocoa futures have their
main weather risk in droughts. Western Africa, especially Ivory Coast
and Ghana, are the largest producers of cocoa in the world. Lack of
sufficient moisture causes the budding pods to wither.
CONCLUSION
The basic tools of operational weather forecasting for the commodity
markets are essential as an invaluable source of information for
traders. All these operations can be reduced to one goal: the best
weather fundamentals for forecasting supply and demand changes.
74
4
Oil and Petroleum Products:
History and Fundamentals
Todd J. Gross
QERI LLC
In this chapter, the conversation on crude oil will be broken into two
main parts. The first section will cover the basics and mechanics of
the current global market, while the second will address historical
price perspective and why the state of the price exists as it does. In
the first section, the basic fundamental and seasonal price drivers of
the new global marketplace for crude oil will be examined.
Subsequently, the chapter identifies the tendencies of crude oil
pricing based upon supply and demand processes that effectuate
seasonal price movements. Some details on the characteristics of
crude oil that can drive price, including quality, grade, location and
transportation, will be next. Finally, the section will conclude with a
discussion of pricing and trading.
The second part will discuss price perspective. It will address how
a US$17/bbl commodity in 2002 could become a US$147/bbl
commodity by only 2008. It will question why the globe always
seems to be running out of oil, while, so far, that fate has yet to be
realised.
WHY OIL?
Critical fuel and elasticity
What can you use crude oil for? This question has a strange, some-
what counterintuitive, answer: not much! However, when crude oil
is delivered to and processed through a refinery, this answer
becomes very different.
75
Crude oil and its products are critical fuels to the world economy
and have huge effects on our daily life. Whether you are using a
plastic cup, filling up your car, heating your home during a cold
winter, or fuelling farm equipment to plant, harvest and bring crops
to market, petroleum plays an important role. The uses of petroleum
products are generally linked to essential modern human needs, and
the demand for crude oil is generally inelastic.
Examples can be too real for those who were waiting in queues in
the aftermath of Hurricane Sandy on the East Coast of the US in
October 2012. Having unfortunately been affected first hand, the
return of 2+ hour queues to fill your car or electric generator,
rationing and police presence at stations resoundingly begs the
inevitable question why dont we just use something else?
Certainly those in New Jersey and New York City would have
instantly shed their place in the queue for a readily available and
cost-beating alternative, but they could not.
There are many reasons for this, most of which point to the factors
of inexpensive cost and infrastructure. Crude oil and its products
have been the least-expensive source of energy across many areas of
the economy for decades. This fact has led to an explosion of
petroleum-related infrastructure that services most daily needs
without a reliable inexpensive alternative. Tankers, refineries,
pipelines, trucks, stations and home furnaces point to a petroleum
infrastructure that makes our society reliant on them while offering
no credible alternative.
These issues infrastructure, price and convenience have caused
a generally limited elasticity of downside demand, which is
supported by the data. As Figure 4.1 shows, the drop-off in
Organisation for Economic Co-operation and Development (OECD)
demand in 200809 was large in absolute terms, but less impressive
in percentage terms: only a 6% decline during the worst recession
since the 1930s. Furthermore, the West Texas Intermediate (WTI) oil
price could barely get back to 2004 levels of approximately
US$50/bbl on a quarterly average basis. This was a level that had
actually not been seen prior to 2004. Such an effect points to a gener-
ally increasing price trajectory since the early 2000s. The elasticity of
demand is roughly a 0.3 ratio to the change in GDP in OECD coun-
tries. Essentially, if the OECD GDP increases by 1%, the demand for
crude should increase by approximately 0.3%.
76
OIL AND PETROLEUM PRODUCTS: HISTORY AND FUNDAMENTALS
Figure 4.1 OECD liquid fuels consumption and WTI crude oil price
Percent change (year-on-year) Price per barrel (real 2010 dollars)
6 150
4 100
2 50
0 0
-2
-4
-6
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
OECD liquid fuels consumption WTI crude oil price
Source: US Energy Information Administration, Thomson Reuters
This phenomenon is a stark contrast to non-OECD growth and

elasticity. It is partially due to the fact that total US demand peaked
in the 200405 time-frame. In Figure 4.2, a much higher elasticity of
demand is indicated for these non-OECD countries. This ratio is
closer to 0.7. With the OECD and non-OECD countries accounting
for about equal amounts of demand, the average elasticity is approx-
imately 0.5.
However, Figure 4.2 shows another important point. Observe the
size and scale of the downturn in the non-OECD during the period
Figure 4.2 Non-OECD liquid fuels consumption and GDP

Percent change (year-on-year)
12
10
-2
-4
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Non-OECD liquid fuels consumption Non-OECD GDP
Source: US Energy Information Administration, IHS Global Insight
77
we focused on in Figure 4.1: the 200809 period. The demand profile

is skewed higher in the non-OECD countries. Growth rates are
higher and the recession area of 2009 is shallower. Does this come as
such a surprise considering Chinese growth rates of nearly 8%, along
with the many emerging economies growing their manufacturing
base? Certainly not; all of these factors lead to limited elasticity of
downward demand for crude oil.
Seasonality
Crude oil and its petroleum products also exhibit significant season-
ality. In Figure 4.3, the monthly demand from 200812 along with the
US Energy Information Administration (EIA) projection for 2013
shows that, even although each year exhibits a different slope
(largely due to macroeconomic developments such as the economic
downturn at the end of 2008), the shape of the demand curves are
nearly the same each year. Basic forces driving oil demand are
approximately the same from year to year. January is plagued by
Figure 4.3 World consumption patterns (200813) (in millions of barrels per day)
92.00
91.00
90.00
89.00
88.00
87.00
86.00
85.00
84.00
83.00
82.00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

World consumption 2008
World consumption 2009 World consumption 2013
World consumption 2010 World consumption average
Source: US Energy Information Agency
78
some refinery turnarounds and holidays, such as the western New

Year and Chinese New Year. Also, it seems like a low level but, as
demand increases on average year on year, the January low
demand is actually on the upswing from the previous Novembers
trough. By February, some refineries return to service around the
globe to meet heating demand in the northern hemisphere. Then,
there is the major second quarter fall off. As spring approaches, the
global refinery complex goes into major turnaround mode.
With major refining regions such as the US taking much of the
refining infrastructure down for maintenance ahead of the
burgeoning summer seasonal usage, along with the moderation of
winter temperatures across the northern tier, demand for petroleum
tends to sag, culminating in the lowest demand period coming in
May. In the second half of the yearly cycle, demand escalates. US
demand for driving and transportation fuel picks up as many take to
the highways for summer vacation. The transportation fuel demand
increase is not only seen in the worlds largest oil consumer but
generally around the globe, culminating in September. The strength
of demand in September is noticeable compared to many other
months. Driving demand is still strong, early pre-winter seasonal
restocking of distillate and heating fuels in Western Europe is afoot
and the global refining industry has yet to go into its autumn mainte-
nance mode. Finally, the waypoint of August and the third quarter
has exhibited stronger demand as air conditioning usage from devel-
oping nations such as Saudi Arabia have kept demand strong while
many are on holiday.
Finally, as can be seen in Figure 4.4, the cycle is complete with the
second refining maintenance season in full swing across the globe as
we enter the fourth quarter. More vacuum distillation units (VDUs)
and atmospheric distillation units (ADUs) are down for maintenance
this time around as opposed to fluid cat cracking units (FCCUs),
which tend to monopolise the spring maintenance season.
Crude grades and locations

Crude oil, when it is taken out of the ground, either offshore,
onshore, using traditional methods or with hydraulic fracturing
(which has precipitated tremendous gains in onshore drilling, espe-
cially in the US), can come with many different chemical make-ups.
Based on the main use for crude oil of refining, each crude grade has
79
Figure 4.4 Seasonal world crude oil consumption

1.00
0.80
0.60
0.40
0.20
0.00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
World crude oil consumptional seasonal 20082012
been tagged with two defining characteristics: light/heavy (based on

the number of carbon atoms) and sweet/sour (depending on the
sulphur content). Grades are then given a name corresponding to
their respective production field name and/or geography, such as
WTI, Brent Blend, Venezuelan Orinoco, Indonesian Minas,
Malaysian Tapis, Saudi Arab Heavy, Oman, Ecuadorian Oriente,
Nigerian Bonny Light and Dubai blends.
The EIA defines light as crudes with an API gravity above 38,
heavy as crudes with an API gravity of 22 or below, medium as those
that fall between 22 and 38 degrees, with 31.1 API as the dividing line
between light and heavy.
According to Platts Energy Glossary:
API gravity = (141.5/specific gravity at 60 degrees F) 131.5.
As for sulphur content, the dividing line is approximately 0.5%

sulphur, where a reading greater than 1.1% sulphur is considered
sour and a reading <0.5% is considered sweet. Figure 4.5 gives the
relationship between most benchmark crudes and where they fall on
the light/heavy, sweet/sour spectrum.
Generally, the process of refining is a chemical process that
enables the breaking and recombining of chemical structures
through heating (as in a VDU or an ADU) or by catalyst (such as an
FCCU), in order to produce petroleum products for commercial use.
Heavy crudes have a higher proportion of large molecules that are
harder to break down, while light crudes have a higher proportion of
smaller molecules. This leads to the lighter crudes that can produce
80
Figure 4.5 Density and sulphur content of selected crude oils

Sulphur content (percent)
Sour 3.5
Mexico Maya
Saudi Arabia Arab Heavy
3.0
Kuwait Kuwait
2.5
United States Mars UAE Dubai

2.0
Iran Iran Heavy Saudi Arabia Arab Light
1.5 Iran Iran Light
FSU Urals
Oman Oman
1.0
Ecuador Oriente
North Sea Brent
0.5 Libya Es Sider United States WTI
Algeria Sahara
Nigeria Bonny Light United States LLS Blend
0.0 Malaysia Tapis
Sweet
20 25 30 35 40 45 50
API gravity (a measure of crude oil density)
Heavy Light
around 40% gasoline versus closer to 20% in the heavier crudes.

These heavier crudes will, in turn, produce heavier distillates to the
tune of around 60% of the barrel. These types of distillates can go to
make heavier materials such as asphalt. In refining, when burning
crudes in a refinery with heavy sulphur content, the output can emit
sulphur dioxide (SO2) or hydrogen sulphide (H2S), a poison gas.
Thus, to meet certain continually stringent sulphur specifications for
petroleum product production, a desulphurisation process has
become increasingly necessary in refineries. These processes help to
refine more sour grades to meet specifications of products such as
diesel, low sulphur diesel or ultra-low sulphur diesel that have spec-
ifications of >500 ppm, <500ppm but >10 ppm and <10 ppm,
respectively (ppm = parts per million). The crude mix has been
moving globally over time towards a heavier (higher) sulphur mix,
which is why the long-term refinery strategy has been to upgrade
their refineries to be able to handle such lower-quality crudes.
Upgraded refineries have been caught between more sweet crudes
and condensates taken out of the ground, and the many global
disruptions in locales such as Libya, which have been throttling
supply.
81
Table 4.1 Monthly Imports from Ecuador to El Segundo, California
Date Company Commodity Entry port State of entry Origin BBLS SULPHUR API
(000s)
August 2012 Chevron USA Crude oil El Segundo, CA California Ecuador 324 1 23.7
August 2012 Chevron USA Crude oil El Segundo, CA California Ecuador 353 2.01 19.4
August 2012 Chevron USA Crude oil El Segundo, CA California Ecuador 374 1 24
82
Furthermore, locations of crudes and their availability have a

huge impact on the type of refinery operations at a specific refinery.
Although most refinery operations are very secretive about their
incoming crude oil slate, some refineries match very well with their
import crude oil.
Let us take Chevron and its assets in the western part of the US,
along with the corresponding crude oil inputs.
Here is a good example. Table 4.1 comes from the EIA website that
tracks company-wide imports on a monthly basis. Why would
Chevron import medium-heavy oil (1924 API degrees) that is
medium-to-high sulphur (12.01% sulphur) from Ecuador to El
Segundo, California? In Figure 4.5, the specs fit Ecuadorian Oriente
Blend well and, considering Chevron has E&P operations in
Ecuador, this would seem to make sense. Ecuador, being on the
western side of South America, has relatively easy transport access to
California, as opposed to the US Atlantic Coast or to many other
places. This crude is a natural fit for California, but why El Segundo?
The answer makes the picture even clearer. Chevron owns and oper-
ates the El Segundo refinery. The reason that the crude is a natural fit
for this refinery is no accident. Chevron intelligently spent quite a bit
of money to upgrade this refinery to the specifications that would
enable it to run such a complex refinery in the state of California (the
most difficult place to refine in the US) and, at the same time, have
the capability to take the medium-heavy, medium-higher sulphur
crude oil of Ecuadorian Oriente.
Locations of major oil supply to demand and limitations of the

transport grid
The worlds oil supply has specific areas of concentration, with many
players moving in and out of prominance over decades. Their rele-
vance is predicated upon their ability to cultivate reserves according
to current economics (as in Venezuela and Iran, Brazil and Angola),
on technology and each countrys willingness to adapt it (as in the
US, with hydraulic fracking) and on their ability to install an infra-
structure that will enable production to grow and hasten its delivery
to market. On the other side of the coin is the ability for the crude oil
being produced to match the corresponding refining capacity. This
match up enables the easy refining of demanded and legally
permitted petroleum products.
83
As we can see in Figures 4.6 and 4.7, production from the Middle
East has historically been sent to the refining areas of the Gulf Coast
of the US, as well as many other refining regions across the world.
However, in response to refining capacity additions and subtractions
by region, oil trade flows have changed. The ability of western
nations to continue to compete globally on refining has become
suspect. The US and Western Europe found their great refining
centres under tremendous duress in 2011. Atlantic Basin crude oil
prices, generally indexed to Brent and imported, were being used as
feedstock to produce higher and higher quality (lower sulphur
content) products. These product specifications were mandated by
the EU and US governments. In addition, these refiners faced ever-
more stringent quality standards on petroleum products while
having to combat a reduction in petroleum product demand since
2008. Furthermore, higher fuel efficiency and the greater acceptance
of clean technologies have also cut into demand.
The business case for continuing to produce petroleum products
in these two jurisdictions if a refinery was lower on the Nelson
Complexity scale and its refining power was generally simple was
becoming increasingly unprofitable. Examples of victims of these
two simultaneously negative phenomena were PetroPlus, the largest
independent refinery in Western Europe, Sunoco, the biggest
refining presence on the US East Coast, and ConocoPhillips, who
divested its downstream business and spun it off into Phillips 66
Figure 4.6 World oil production by region (millions of barrels per day)
Asia Pacific, 8.1 North America

14.3
Africa, 8.8
South & Central
America, 7.4
Europe & Eurasia

Middle East, 27.7
17.3
Source: BP Statistical Review, 2012
84
(with this last case probably augmenting efficiencies in manage-

ment). The replacements for these refineries were generally Indian,
Chinese and Middle Eastern refining capacity additions. With fewer
regulations for start-up, cheaper labour (none or fewer union labour
constraints) and closer proximity to the source of crude oil, the
trends depicted in Figure 4.7 are not only set to continue, but to be
accentuated in the coming years. After all, no major refineries have
been built in the US since 1976. Capacity grew solely through
upgrades and increases in complexity. Meanwhile, Asian refiners,
less hindered by regulation and clean air rules, have been building
brand new, more efficient refineries that can actually fill the gaps left
by those archaic, decommissioned refineries of the west.
However, as this all seemed to be speeding out of control in the US
and Western Europe, a new technological breakthrough came along
just in time to give at least a temporary reprieve for many US East
Coast refineries. Domestic crude oil, produced through techniques of
hydraulic fracturing from the interior of the US, has made its way
across the country to the refining centres in a cost-efficient way. This
trend has given some of these refineries hope. We will talk more
about the phenomenon later in this chapter but, with the Carlyle
Group purchasing part of the Girard Point refinery from Sunoco and
Delta Airlines purchasing the Conoco Phillips Trainer, PA refinery,
the progression of this global change in regional refinery economics
Figure 4.7 Refining capacity market share evolution
2001 global capacity: 83.4mb/d 2011 global capacity: 93.0mb/d

22.99%
26.22% 24.21% 31.33%
7.09%
3.80% 7.49%
3.57%
8.09%
8.61%
30.19% 26.42%
North America Middle East

South & Central America Africa
Europe & Eurasia Asia Pacifc
Source: BP Statistical Review, 2012
85
seems to have been stayed. Additionally, the US Gulf Coast has been
importing less and less crude to run its refineries, and within 23
years the region should not need to import crude oil at all.1
The future trajectories of production, refining and storage are
beginning to change the market and may slowly change how crude
oil and petroleum are priced. Despite the extraordinary growth in US
production, there is little risk to the status of the Middle East and
former Soviet Union (FSU) as major producers. The inclusion of
Iraqs huge reserves under a market-oriented and ambitious regime
provides optimism for the continued strength of Middle Eastern
crude oil production. With Brazil and Russia having gained in
economic prominence, these countries will also have the ability to
marshal larger resources toward oil exploration and production
(E&P) for increasing contributions in the global crude oil supply mix.
Also, new technology and high prices have encouraged a renais-
sance in US oil production.
The real issues that have arisen from this sea change are logistical.
How does the crude get from production areas to the refiners, and
what are the risks along these routes? The flow of crude oil from
Middle Eastern countries to jurisdictions East of Suez has been
growing for decades. However, with more and more of global oil
production heading in this direction, the world oil transit choke-
Figure 4.8 Refinery additions (201035)

5
20302035
20252030
4 20202025
Millions of barrels/day
20152020
20112015
3
0
US & Latin Africa Europe FSU Middle China Other
Canada America East Asia
Source: OPEC World Oil Outlook, 2011
86
points should be examined, especially in light of the risks concerning

the Strait of Hormuz.
Crude transport and chokepoints2

There are seven major world transport and chokepoints for crude oil
tanker movements (see Figure 4.9):
The Strait of Hormuz;

The Strait of Malacca;
Bab el Mandeb;
Turkish Straits;
Danish Straits;
The Suez Canal/SUMED Pipeline; and
per day in transit is shown in Table 4.2.

IA, about half of the worlds oil production
routes, the rest mainly transits through
he Strait of Hormuz and the Strait of Malacca
d Pacific Oceans are by far the most strategic.
Figure 4.9 Potential chokepoints to global crude transport
87
Table 4.2 Volume of crude oil and petroleum products transported through world
chokepoints (200711)
Location 2007 2008 2009 2010 2011
Bab el Mandeb 4.6 4.5 2.9 2.7 3.4

Turkish Straits 2.7 2.7 2.8 2.9 N/A
Danish Straits 3.2 2.8 3.0 3.0 N/A
Strait of Hormuz 16.7 17.5 15.7 15.9 17.0
Panama Canal 0.7 0.7 0.8 0.7 0.8
Crude oil 0.1 0.2 0.2 0.1 0.1
Petroleum products 0.6 0.6 0.6 0.6 0.6
Suez Canal and SUMED Pipeline 4.7 4.6 3.0 3.1 3.8
Suez Crude Oil 1.3 1.2 0.6 0.7 0.8
Suez Petroleum Products 1.1 1.3 1.3 1.3 1.4
SUMED Crude Oil 2.4 2.1 1.2 1.1 1.7
Notes: All estimates are in million barrels per day. N/A is not available. The table does not
include a breakout of crude oil and petroleum products for most chokepoints because only the
Panama Canal and Suez Canal have official data to confirm breakout numbers. Adding crude
oil and petroleum products may be different than the total because of rounding. Data for
Panama Canal is by fiscal year.
Source: EIA estimates based on APEX Tanker Data (Lloyds Maritime Intelligence Unit);
Panama Canal Authority and Suez Canal Authority, converted with EIA conversion factors
Let us talk about the granddaddy of them all at first, the Strait of
Hormuz, which is located between Oman and Iran and connects the
Persian Gulf with the Arabian Sea. Here, roughly 35% of all seaborne
traded oil and 20% of all oil traded worldwide passes through on a
daily basis. More than 85% of these crude oil exports go to Asian
markets such as Japan, India, South Korea and China. At the
narrowest point, the Strait is 21 miles wide and the width of the ship-
ping lane in either direction is only two miles, separated by a
two-mile buffer zone. The alternatives are woefully inadequate.
Pipeline replacement capacity currently only offers 45 million
barrels a day of unused capacity, and trucking would add only a
maximum of a few hundred thousand barrels per day. Most tankers
going through the Strait of Hormuz run greater than 150,000 dead-
weight tonnage (DWT) these are very large tankers. A block of the
Strait of Hormuz would result in a shortfall of undelivered crude oil
of perhaps up to 12 million barrels a day.
The Strait of Malacca is the other main strategic point. It is
located between Indonesia, Malaysia and Singapore (where the big
Pulau Bukom 500,000 barrel-a-day Shell refinery operates), and
links the Indian Ocean with the South China Sea and Pacific Ocean.
88
This is the key chokepoint in Asia. With 13.8 million barrels per day
(mbpd) flowing in 2007, the Strait had ratcheted flows up to an esti-
mated 15.2 mbpd in 2011. At the narrowest point, in the Phillips
Channel of the Singapore Strait, Malacca is only 1.7 miles wide. If
the Strait was blocked, nearly half of the worlds fleet would have
to reroute around Indonesia. With so much crude flowing through
this waterway, it would not go undelivered as in the case of a
blockage of the Strait of Hormuz; it would just have to be rerouted
at greater costs and time to market.
The rest have their strategic interests too. The Turkish Straits are
important for western pricing because it is a main thoroughfare that
transports Russian crude exports, as well as exports from Azerbaijan
and Kazakhstan, to Western European refineries. Weather often
impacts transit in winter, forcing additional transit time of up to
weeks in some cases. Finally, a Bab el Mandeb closure could keep
Persian Gulf tankers from reaching the Suez Canal as it is located
between Yemen, Djibouti and Eritrea, and connects the Red Sea with
the Gulf of Aden and the Arabian Sea. Most transit goes north to
destinations in Europe, US and Asia. If impassible, it would redirect
3.4 mbpd around the southern tip of Africa, a significant addition of
transit time.
Crude pricing and trading

Not all crude oil that is produced and delivered goes directly into the
refinery for processing. The crude that awaits refining in any time-
frame is held in storage. Storage is the most significant statistic of
over- or under-supply in the crude oil market.
Most notable has been the effects of storage levels and capacity in
Cushing, OK, the delivery point of the CME/Nymex WTI crude oil
futures contract. Being a landlocked area with limited capacity and
limited transit to and from the storage tanks, the Cushing phenom-
enon played a major role in the term structure of the Nymex futures
contract through 2010. As one can see from Figure 4.10, a significant
amount of storage capacity has been added to Cushing inventories
since the third quarter of 2010. This fact has alleviated some of the
risks of storage congestion and stock-out phenomenon that has
plagued this storage area, and therefore the Nymex pricing of
prompt/term spread relationships.
When storage levels approached high percentages of capacity, the
89
Figure 4.10 Cushing storage

80
Shell capacity
70
60
Million barrels
50
40
Working
30 storage
capacity
20 Inventories
10
0
Sep 30, 2010 Mar 31, 2011 Sep 30, 2011 Mar 31, 2012
WTI market would go into a strong contango. Known as storage

congestion, this has not been overly studied in the theory of
storage and academic circles. The volatility in the markets is usually
greatest when the market is near stock-out capacity, and when there
is not a credible alternative to satisfy demand (Kaldor, 1939;
Working, 1948, 1949). The opposite is also true. For certain commodi-
ties, where storage is not universal and is limited, and there is little
outlet for continued production, the price will pick up significant
volatility as full storage is approached. Spot prices will tend to
become more volatile when storage operators are not seasonally
involved in the market or their facilities are near capacity (in a similar
fashion to how front-month futures price volatility tends to increase
as expiration draws near (Samuelson, 1965)). As prices plummet and
the market becomes more volatile, the percentage movement in
underlying pricing and spreads can rival even the most extreme
stock-out scenarios. Research from the likes of Carlson, Khokher,
and Titman (2007) and Evans and Guthrie (2009) has suggested that
there is more of a U-shaped relationship between spot price volatility
and the slope of the term structure of forward prices. Strangely, both
phenomena are less likely with greater storage capacity!
We can see in Figure 4.11 the takeaway points for Cushing crude oil
and many additions to alleviate the so-called bottlenecks that inhibit
the transit of crude oil to the Gulf of Mexicos major refining area have
been, and continue to be, implemented. However, generally, there
are inflows from local production and the incoming crudes off of the
90
Enbridge pipeline into the Cushing area. Outflows had been going to
the only consumers on the block, the local refineries. Other pipeline
capacity is in full swing, such as the Seaway pipeline reversal (it used
to bring crude up from the US Gulf Coast to the PADD II refineries
until domestic Bakken and Southern Canadian production exploded),
alleviating most bottlenecks. The consuming pipelines name the
destination. BP is flowing towards Chicago (or, more specifically,
Whiting, Indiana) to its 410,000 bbl/day Whiting refinery. Likewise,
the Ponca line heads to the Phillips 66, Ponca City refinery (at 195,000
barrels/day) and the Ozark pipeline takes off to St Louis area to
supply the Phillips 66 (formerly Conoco Phillips) Wood River
Refinery at (300,000+ barrels/day). Also worth noting is CVR
Energys 115,000 barrels/day Coffeyville, KS refinery, which has its
own line coming from the Oil Hub. Then there are other inputs, such
as the Enbridge Spearhead pipeline that delivers more Canadian
crude to Cushing.
Furthermore, Figure 4.11 shows existing and proposed pipeline
expansions, which continue to address transport issues of crude oil
Figure 4.11 North American oil pipelines
Sources: Map from Canadian Association of Petroleum Producers. TransCanada overlay

from TransCanada Corp. Assembled for Watershed Sentinel by Arthur Caldicott.
91
from production areas in the north to the refining centres in the Gulf
Coast.
Finally, Figure 4.12 points out the more important figures for
pricing the WTI near term structure. Two very different states of the
world existed for prompt second-month WTI spreads in September
2008, and then quite the exact opposite in January 2009. On
September 13, 2008, the 110 mph Hurricane Ike crashed into the
Houston Gulf Coast, delaying crude oil imports and disrupting
infrastructure up the Houston Ship Channel and the Loop, to the
point that Cushing inventories plummeted and the spike in
prompt/second WTI spread blew out to US$29/bbl on expiration
(See Figure 4.14). Then, only four months later, the opposite was
true. Crude oil inventories were approaching a limit at 80% of then
storage capacity at Cushing, Oklahoma, and global inventories were
dramatically swelling. With capacity at just under 40 million barrels
in early 2009, the inventories ballooned to just under 35 million
Figure 4.12 Continuous prompt/second nearby spread WTI

Daily CL CL spread 3/13/20074/23/2012 (NYC) Price
USD
Line, CL CL spread, trade price (last)
Bbl
12/13/2012, -0.52, +0.02, (+3.70%) 10.5
10
9.5
9
8.5
8
7.5
7
6.5
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
-4.5
-5
-5.5
-6
-6.5
-7
-7.5
-8
.12
A M J J A S O N D J F MA M J J A S O N D J F MA M J J A S O N D J F MA M J J A S O N D J F MA M J J A S O N D J F MA M J J A S O N D J F MA
Q2 07 Q3 07 Q4 07 Q1 08 Q2 08 Q3 08 Q4 08 Q1 09 Q2 09 Q3 09 Q4 09 Q1 10 Q2 10 Q3 10 Q4 10 Q1 11 Q2 11 Q3 11 Q4 11 Q1 12 Q2 12 Q3 12 Q4 12 Q1 13
Source: Thomson Reuters
92
Figure 4.13 Crude oil stocks: Cushing, OK

40000
35000
30000
25000
20000
15000
10000
5000
0
0 4 08
08
08
08
08
04 08
09
04 09
9
00
00
00
ov 200
00
00
00
Fe , 20
20
20
Se , 20
Fe , 20
0
,2
,2
,2
,2
,2
,2
,2
,2
,2
4,
4,
,
04
04
04
04
04
04
04
04
04
04
r0
l0
n
ar
ay
ct
ec
ar
Ju
Ap
Au
Ja
Ju
Ja
O
M
M
M
D
N
barrels (Figure 4.13). The ensuing change in prompt second-month

spread was dramatic.
In the 2000s, the small glimmer of technological advancement in
hydraulic fracturing almost entirely captured headlines in the
natural gas arena as a production game changer. The realistic impact
on global crude supplies was initally discounted because, although
technology made additional US production theoretically possible,
the barrels could not be moved from these interior locations due to
the lack of midstream infrastructure. Pipeline assets usually take
onshore crude oil from E&P areas to refinery gates for easy loading
into the facility to make product. The completion of many new assets
to fulfill additional takeaway-capacity needs seemed several years
away. There was a trend towards the bankruptcy of East Coast
refineries that had similar issues to that of their cousins in Europe,
and the stranded nature of this new crude productions location. The
only outlet seemed to be to get the oil to Cushing, OK. However,
events have begun to change this state of affairs, although a lot more
needs to be done to rectify the two main dislocations in the oil and
products markets that are inexorably intertwined: the Brent/WTI
pricing mechanism and the East Coast refinery market.
The Brent/WTI spread market flourished because of two very
important attributes, that were both financial and physical in nature.
93
Figure 4.14 WTI Brent price spread (January 2013 contract)

Daily WTCL-LCOF3 7/13/200712/18/2012 (LON)
Price
Line, WTCL-LCOF3, trade price (last) USD
12/13/2012, -22.32, +0.41, (+1.80%) Bbl
3
2
1
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
-11
-12
-13
-14
-15
-16
-17
-18
-19
-20
-21
-22
-23
-24
-25
-26
-27
.12
Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4
2007 2007 2009 2010 2011 2012
Source: Thomson Reuters
The flagship contract on the Nymex was WTI, which fostered an

active, entrepreneurial place for hundreds of traders to provide
liquidity for the crude oil futures contract. There was always a trans-
parent price that could be transacted. Likewise, the counterpart in
the UK was the Brent Blend contract, which started on the IPE before
being owned by the ICE. The Brent contract had a little different
make-up. Although not as liquid as a futures contract, it had the
unique characteristic of being directly tied to Dated Brent, a more
commonly used benchmark for the spot price of crude oil.
Furthermore, the the two contracts were easily linked as there was a
direct route to get Brent to the same place as WTI. Take a loading on
a tanker in the North Sea and drop it off at the Loop, the Louisiana
Offshore Oil platform, or in the Houston Ship Channel. The crude
could then be piped onshore and up yet another pipeline, sending it
north up to the Midcontinent and Cushing, OK (one of these
pipelines being Seaway). Brent typically traded at a discount to WTI,
because most incremental refining barrels were absorbed by the US
refining machine and therefore Brent traded at a discount to Light
94
Louisiana Sweet (LLS), generally by the cost of transportation to the

US Gulf Coast (USGC). LLS and WTI were generally linked by inex-
pensive pipeline economics that would bring crude from the Gulf via
the pipeline. There existed a very liquid market in buying and selling
Brent cargoes, hedging with more liquid WTI futures and trading the
spread back and forth actively.
The spread (and its economics) were severely disjointed by the
unexpected explosion in Midcontinent supply coming from Bakken
shale in the US and production from Southern Canada. This increase
came at the same time that production in the North Sea was
declining to a point that there was a noticable drop in production out
of the Brent, Forties, Oseberg and Ekofisk (BFOE) cocktail.
Specifically, Nexens Buzzard field of 200,000 barrels per day,
approximately 10% of the North Sea production, has had major,
continuing maintenance problems. The BFOE cocktail that cargoes
were priced off of had compounding issues when Buzzard was
down. Buzzard Forties production, one of the lowest-quality crudes
in the cocktail, had a knock-on effect on price. As production went
down, supply would shrink drastically. In addition, the cheapest
element of the cocktail was diminished, leaving even more expensive
crudes to make up the price. The cocktail had been priced on the
cheapest-to-deliver crude. This phenomenon adds extra elasticity to
the Brent/WTI movement that had come to plague the market.
Meanwhile, many remedies were being sought to alleviate this
price differential on Brent/WTI. While the East Coast and Gulf Coast
refineries were having their crude feedstock priced off of Brent, the
Midcontinent PADD II refineries enjoyed the economics of land-
locked Cushing pricing. There was financial incentive to redistribute
the crude and try to alleviate bottlenecks. The sale of the Seaway
pipeline and the reversal of flow has begun to help, but much more
needed to be done. The Brent/WTI price differential (still at Brent
US$20 over) did not look like it would relent soon. The reversal of
Seaway, which has removed 150,000 bbls/day along with another
250,000 bbls of throughput to be added in 2013, should help. The
political football of the Keystone XL pipeline, as well as many lesser-
known avenues, had been put to work to alleviate the glut and to
take advantage of the US$20+ price differential. However, unfore-
seen issues such as the limited storage tank capacity at Jones Creek,
Texas, had diminished the Seaway pipelines effectiveness. This final
95
outlet for the Seaway pipeline only boasts 2.6 million barrels of
storage and has limited the ability of Seaway to move all of its
400,000 bbls/day capacity to the Gulf. These constraints had
prevented a resolution of the spread relationships.
With many alternatives for crude transport unavailable, the
markets turned to an old school form of crude oil transportation: rail-
roads. Rail loadings of oil have been soaring and the economics make
sense. With many new terminals being built to handle much of the
throughput, the transport of crude via rail has been able to alleviate
some of the issues. This solution has changed the equation enough to
rationalise the economics of two East Coast refineries. With the hope
of getting Midwest crude oil, the business case has changed from an
unprofitable venture such as those in Europe to big opportunities for
those including Monroe Refining (a division of Delta Airlines) and
the Carlyle Group. Both investors have bought two main East Coast
refineries previously set for closure because of poor economics. The
ability to receive shale crude oil as feedstock has helped to make the
business case to keep these refineries open. Furthermore, in Monroe
Energys case, their supply chain of jet fuel in the New York market
and ability to supply competitors makes it a sound investment, with
some personnel who used to work at the refinery already being part
of the Monroe team.
Figure 4.15 Oil on rail transport

64,663
51,482
36,544
26,247
16,789
11,389
11,324
10,843
8,583
6,784
3,395
2,860
2,832
2,650
2,498
2009 2010 2011 2012

Source: AAR Weekly Railroad Traffic
96
Figure 4.16 North Dakota railroad map
Source: www.Trainsmag.com (May 2012)
As Figure 4.15 shows, there has been a jump in railcar loadings

with petroleum. With growth near 45% for 2012, rail movement of
crude oil is showing itself as the stopgap measure of choice between
the production and demand today and the time when lower trans-
mission cost pipelines are built. Economics are showing the railing of
crude from Bakken to the Gulf Coast as an approximate mid teens
per barrel cost. These economics have enabled such railing. Risks to
this method have been highlighted with the crude oil rail tanker acci-
dent July 6, 2013 in Lac-Megantic, Quebec.
Crude markets and trading

Oil is traded physically in many corners of the globe. With bench-
marks such as Dated Brent off the BFOE pricing, the Japanese Crude
Cocktail (JCC) pricing many far eastern contracts, Oman/Dubai
pricing a lot of the Middle East sour crudes and FSU Urals pricing
much of Eastern Europe and distillate products, these crude oil
97
benchmarks are all interrelated. Global pricing is influenced by what

refining capacity is operating or down for maintenance, and gener-
ally if there are problems in loading (for example, in Nigeria when
there are militant attacks). Maintenance in fields such as the Buzzard
field in the North Sea can also have an outsized influence on these
benchmarks. Grades are crucial and which refineries take those
grades can make the difference between a wide or tight sweet/sour
spread differential. Crude oil is mainly moved on Dirty tankers with
>150,000 DWT, or very large crude carriers (VLCC).
Specifically for crude oil, Dated Brent is the most widely accepted
global crude oil benchmark, and always faces intense scrutiny from
producers, end-users and regulators. Dated Brent is generally used
as a sweet crude benchmark and prices crude in the North Sea, West
Africa, the Mediterranean, South and Latin America, Canada,
Central Asia and Russia. More than 60% of the worlds internation-
ally traded crude oil is priced against Dated Brent.3 Dated Brent is
the price assessment of physical cargoes of North Sea light sweet
crude oil. The term Dated refers to the physical cargo price for
North Sea Brent light crude which has been allocated a specific
forward loading date (1025 days ahead). The North Sea light sweet
crude oil grades Forties, Oseberg and Ekofisk are also deliverable
against the Dated Brent contract known as alternative delivery, as
the combination of all four crudes is known as BFOE. This combina-
tion gives Dated Brent a supply of approximately 1.4 mbpd and
provides enough liquidity to sustain it as a benchmark. The window
for pricing Brent occurs at 4:30 pm, London time. When prices of
Dated Brent are high, the North Sea attracts crudes from West
African and the Mediterranean, while when the benchmark price is
low, North Sea pushes crudes to other places, such as the US Gulf
Coast. Historically, Malaysian Tapis and Indonesian Minas had been
the benchmarks for sweet crude in the Far East and Asia. However,
with production becoming smaller and smaller and fewer barrels
being available for export, the region has turned to Dated Brent for
much of its pricing, with even Indonesia pricing its barrels off this
global benchmark. The Asian version of Dated Brent is priced off a
Singapore pricing window at close of business 4:30 pm, Singapore
time.
As for sour crude oil, Dubai had historically been an Eastern
benchmark. However, as physical export supplies became scarce in
98
the Asian/Pacific regions and import demand climbed to levels

greater than 17 million barrels per day, these benchmarks needed a
little help. Therefore, the Dubai benchmark has added Oman and the
Dubai Mercantile Exchange has touted its Oman futures contract
that has a delivery point east of Suez and 860,000 barrels/day of
export volume. As for the pricing of many grades of crude for export,
these benchmarks enable pricing schemes, but vary based upon
destination. For example, Saudi Arabia may price exports to Europe
based upon the Brent Weighted Average (BWAVE) price, its exports
to Asia based upon Oman and Dubai and its exports to the USGC
based upon Argus Sour Crude Index (ASCI), an index of delivered
sour crude to the USGC.
Product points have just as much relevance. With product trade
and transport becoming more of the global petroleum trading
market through the 2010s (International Energy Agency, IEA, 2012),
one has to be cognisant of those that produce and those that will
receive. The ports of interest are mainly the USAC, USGC, Sullom
Voe terminal in the UK, Amsterdam, Rotterdam, Antwerp (ARA),
Ras Tanura in Saudi Arabia, Singapore, Chiba in Japan, Shanghai
and the MED terminals in Fos Lavera near Marseille, France. Most
product pricing hubs are aligned with an important maritime port,
usually one or many large refineries and, of course, most impor-
tantly, storage facilities for petroleum products.
Most petroleum products are moved on barges or clean tankers of
around 60,000 DWT. The terms FOB (free on board) and CIF (cost,
insurance and freight) denote whether the pricing is based on the
buyer providing transport and the seller delivering the barrels on
board, or the seller covering transportation and insurance costs to
deliver the cargo to the buyers destination port.
On a global basis, the trend is for less refining activity out of
Western Europe and for those losses to be supplanted by gains in
India and China. The growth of the giant Jamnagar complex in India,
along with the upgrade of the Essar Oils refining complex from
300,000 to 600,000 barrels per day, has shown Indias high-profile
strength in the refining sector. China has added a multitude of
refineries since the late 2000s. These refineries have been located in
many different areas of China, and although not aggregated into a
single massive refining complex, they represent huge additions in
refining capacity.
99
As this was taking place, the bankrupcy proceedings of PetroPlus

mainland European holdings was happening. These older, less
complex refineries, which originate crude oil from long distances,
and which are forced to deal with organised labour issues, have
become less competitive on the global landscape. The final straw was
the recent Libyan revolution that took away the much-needed sweet
crudes that some southern European refineries had used for feed-
stock, not easily replaced by the sour FSU Urals blend that was the
most readily available swing supply at the time.
As refining moves East, pricing and benchmarks for the worlds
refineries will change. Figure 4.17 shows many of the benchmarks for
crude oil and the pricing points. As the North Sea faces continuing
decline in output capacity, the US production pushes Nigerian and
Angolan crudes to the East, and refining interests procure more
marginal barrels from Middle East sources, some of the refining
benchmarks may move towards the Oman contracts on the Dubai
Mercantile Exchange.
The US is a different matter. With its strong refining base in the
USGC, its excess capacity has been mobilised to export products to
certain markets, many located in South America. As South American
demand for products has continued to climb, along with the closure
of the Hovensa refinery on St Croix, the Valero Aruba refinery and
the chronic maintenance needed at the giant Paraguana refining
Figure 4.17 Pricing benchmarks for global crude oil
North Sea - Brent
FSU - Urals
United States - WTI Algeria - Sahara Blend

United States - LLS
United States - Mars Libya - Es Sider
Mexico - Maya
Ecuador - Oriente Nigeria - Malaysia - Tapis

Bonny Light
Kuwait - Kuwait IranIran

- Iran Heavy
- Iran Light
Saudi Aradia - Arab UAE - Dubai
Heavy
Saudi Aradia - Arab Oman - Oman
Light
Source: US Energy Information Administration
100
complex in Venezuela, US Gulf Coast refineries have been recruited

to export products to meet demand in the south.
Several commodity exchanges offer futures contracts, some which
may be settled by physical delivery of the underlying crude or
product, and some that may be settled financially. These futures are
widely used by producers, refiners and large consumers of crude
and products for price risk management, and are also traded by
speculators and investors who desire exposure to energy prices. The
main futures and options markets are traded on the CME and the
Intercontinental Exchange. The products listed are WTI, Brent, Ultra-
Low Sulfur Distillate (which was previously known as Heating Oil),
Gasoil, RBOB and many other locational contracts (such as flat-
priced delivery points of USGC, ARA or Singapore) that are listed on
ICE or cleared on CME Clearport. Although Heating Oil and Gasoil
have been the mainstay for pricing of global distillate demand since
the early 1980s, these contracts are slowly being replaced by their
lower sulphur counterparts that are becoming a larger segment of
distillate demand, with the ICE and CME adding Low Sulphur
Gasoil contracts since 2012.
A HISTORICAL PRICE PERSPECTIVE

Figure 4.18 illustrates a historical perspective of oil prices and some
of the major effects since the early 1970s. The first commercially
drilled oil well was drilled near Titusville, Pennsylvania, in 1859 by
Edwin Drake. Even although kerosene production from crude oil
goes back to the Babylonians uses of petroleum, the implementation
of the combustion engine and later uses in transportation were the
main drivers of the pursuit of crude oil production. Early on in petro-
leum history, 90% of the worlds crude oil was in Baku, Russia, and
after a century and a half Russia has once again become the largest
producer of crude oil, but, according to the IEA report of October
2012, by 2017 the US will resume its place as the worlds largest oil
producer.
However, the history of modern oil pricing really started in 1960
with the birth of Organization of Petroleum Exporting Countries
(OPEC). During that era, Western demand for oil was mostly met by
international oil companies (IOC) and production was mandated by
quotas set by the Texas Railroad Commission. What followed was a
series of events that turned the price and availability of oil upside
101
Figure 4.18 Oil disruptions, OPEC spare capacity and crude prices
25% Market fears of

Threatened oil supply an Iran-related
Disrupted oil supply Hormuz
Spare capacity (EIA) disruption
Crude oil price (RHS) US$120
faded after
20% April
Share of world oil demand
US$100
Iran
Current US$ per barrel

15%
US$80
US$60
10%
IranIraq war
US$40
5%
Iran
revolution US$20
Arab oil embargo
Gulf war I Gulf war II
0% US$0
70
72
74
76
78
80
82
84
86
88
90
92
94
96
98
00
02
04
06
08
10
12
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
VZ 0203, Iraq 03, Nigeria 03>, Libya 10>, and others
Source: The Rapidan Group
102
Prices: 7273 Arab Light, 74present US refiner average imported crude cost.
down. Back in 1956, M. King Hubberts presentation to the American

Petroleum Institute suggested a peak in US production that actually
took place (albeit for the time being) in 1970. Then, in March 1972, the
Texas Railroad Commission declared that having quota restrictions
on production was not necessary because demand had outstripped
supply and that producers could produce at their capacities. This
event ushered in the shift of power over global pricing of crude oil
from the West to OPEC. Shortly thereafter, there was the Yom
Kippur War, and, with the Wests support of Israel, the ensuing Arab
oil embargo that lasted from October 1973 to March of 1974. Prices
skyrocketed.
Once again, Hubberts ideas of a global production peak had
permeated into the market, now pointing to global production
peaking around 1995. With the growth in production coming from
the Middle East, and the economic changes and expansions on the
horizon set for what would become the nations of the G7 and eventu-
ally the G20, the secular movement of Western economic powers
taking from the Eastern producers became the emerging status quo.
In 1979, the Iranian revolution added another jolt to the spot crude
oil price. In late 1978, a strike by foreign workers who later fled the
country during the 1979 revolution, helped Iranian production
decline from more than six million barrels a day from which the
production has yet to recover. The 1979 revolution led into the 1980
Iran/Iraq war, signalling a second oil price spike in a decade.
However, with the resurgence of Soviet Era assertion for energy
dominance and new technologies for exploration and production,
first the USSR and then Saudi Arabia in the 1990s stepped in to fill
the gap to become the number one and number two global oil
producers. With the emergence of a general global peace, excess
supply and better technology, Hubberts predictions looked implau-
sible. In fact, in the late 1980s there were several events that helped to
vault prices lower. In 1986, with Mexico becoming a strong regional
player in the West, the Mexican government offered to price crude
delivered on a netback basis. This meant that they would price crude
oil based upon the price at which products could be sold. As refiners
were basically guaranteed profits, they produced until there was
major oversupply, which pushed WTI prices on Nymex to
US$9.75/bbl (meaning that prices were down by 80% in a few short
years). The lower price regime continued generally through the late
103
Figure 4.19 Crude oil production trends (since 1960)

14
Former USSR
12
United
10 States
Million barrels per day
Saudi
Arabia
8
6
Russia
4
Iran
2
0
1960 1970 1980 1990 2000
1980s. A perfect example was a pre-OPEC meeting headline in The

New York Times Business Section in November 1988, which read:
Three Cheers for US$5 Oil. At the time, Kuwait was a chronic over-
producer and kept the prices down. The Saudis suggested that they
could just flood the market with oil and be the last one standing...at
US$5/bbl. This particular dynamic seemed to replay over the next
few years as a recurring theme, even although OPEC was able to
come out with an agreement in November 1988.
Kuwaits overproduction was not such a black and white case.
During this era, OPEC quotas were actually important. They were
hard to enforce, but markets did enforce them, as otherwise the price
would plummet, and OPEC ministers were forced to act. Kuwait
was coming into its own at that time in oil production. The country
was able to invest in production and grow its production capability,
but they wanted to sell this new capacity. These aspirations eventu-
ally caught the ire of Saddam Hussein and Iraq. The Iraqis believed
that Kuwait was originally (and still was) the 18th province of Iraq,
and, due to its chronic overproduction, Kuwait was blamed for
keeping oil prices low.
Much to the disbelief of the West (even although Iraq had amassed
hundreds of thousands of troops on the border days before), Iraq
104
invaded Kuwait on August 2, 1990 and away went the price of

crude oil. The price peaked in October of 1990 and, when the US lead
a coalition to free Kuwait in January 1991, it ushered in an era of a
stronger Western military presence in the Gulf region, relatively
unchecked after the break up of the Soviet Union in 1989. The result
was a stable environment for oil prices throughout the 1990s. Excess
OPEC capacity trended higher throughout the decade as OPEC
nations added capacity faster than demand, and this excess capacity
reached a level not seen since the Iran/Iraq war.
In 1998, with the price of WTI trading near US$10/barrel, there
was once again trouble in OPEC. The supply situation had placed
Sunni-lead Saudi Arabia at loggerheads with Shiite Iran based on
pumping. By default, Saudi Arabia had become the swing oil
producer in times of market shortfalls, tightening their new alliances
with the consuming nations in the West they had become the de
facto central bank of oil. With overproduction coming from Iran and
Venezuela, the balances were once again hard to maintain. The
market found a bottom, but not until a real resolution on production
and quotas were reached by these countries.
This market downdraft was not without casualties. With
Hubberts predictions about 1995 all but forgotten, perhaps the best
trade of the decade happened with oil near US$10/bbl. Exxon
bought its largest rival, Mobil, in 1999 at the bottom of the market.
Hubberts global production assessments were not off, but some-
what delayed by the one thing that has also reemerged in the
previous decade: technology. The ability to leverage existing oil
fields by pumping large amounts of water into a field and thus
expanding its production capacity, enabled big oil fields, such as the
Ghawar oil field in Saudi Arabia, to increase or sustain its production
capability when it should have begun to decline. Saudi Aramco,
boasting the best technology of any oil company in the world, was
defying production constraints with new technology.
Finally, around the new millenium, some old predictions began to
take hold. After the economic downdraft in 2002 precipitated by
September 11th and the South American debt crisis, the growth of
emerging economies became noticeable. The Brazil, Russia, India
and China (BRIC) economies began to grow to a point where the
consumption of crude oil and refined products were overwhelm-
ingly dependent on the ability to find oil.
105
The era of finding onshore super-giant oil fields was gone. The
new cost of finding fields and extracting was increasingly being
focused on deep-water offshore finds that were expensive and risky
to excavate. With very large fields such as Mexicos Cantarell in
decline, and the West feeling the pinch of the fall-off in production
from the Hugo Chavez regime in Venezuela, there was great concern
in the race for the marginal barrel. Areas such as the North Sea had
begun a decline that continues to the present day. The one major
bright spot that Figure 4.19 does not point out is the upswing in US
production that now boasts greater than 7 mbpd, reversing the
downward trend which was intact since 1970.
Let us now look at Figure 4.20, which illustrates the growth in
consumption of the largest driver of the decade, China. Amazingly,
since China became a net importer of crude oil, its shortfall has
grown substantially to make it the second largest consumer of crude
oil after the US. This rapid growth and migration of the populace to a
middle class that is a global consumer of crude oil products has had
a profound effect on price and excess capacity (as shown in Table 3).
According to EIA projections, this trend will continue going
forward through to 2035. With much of the future growth in liquids
consumption coming from China, India, other non-OECD Asia and
the Middle East, much of the supply growth will have to come from
somewhere. Interestingly, OPEC is showing a growth in market
share from about 40% to 42%. Therefore, the promise of Iraqi growth
may have some lasting effects on keeping OPEC share growing.
Meanwhile, as shown in Table 4.4, with the production declines in
the OECD countries, the lone shining star is the US thanks to the
shale production boom that may even supercede the estimates which
may crowd out some OPEC production growth. The IEA claims that
the US will be the worlds largest oil producer by 2017. This implies a
staggering growth rate, which may be difficult to achieve given the
typically high decline rates for most new wells in the Bakken and
Eagle Ford shales.
There are a few things to note based upon the overall trends.
Looking once again at Figure 4.18, the tightness of the supply
demand balance that ushered in this new era of prices largely took
effect when the excess OPEC capacity shrank back below 3% of
global production (about 2.7 mbpd). At the same time, there was a
second stage ramp-up in Chinese demand (as shown in Figure 4.20)
106
Figure 4.20 Chinese net oil consumption

Thousand barrels per day Forecast
12,000
10,000
Consumption
8,000
Net imports
6,000
4,000
Production
2,000
0
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10
11
12
13
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
20
20
at the acceleration point around 2003. Thus, the new price regime
entered the markets. With similar shortages during the first Gulf
war, the nominal price reached US$41 in late 1990. Contrast that time
with early 2009, in an oversupplied environment of having 6%+
excess capacity the price was only able to fall to US$32/bbl. This
price action speaks to a new price regime.
Note the price assumptions listed by the EIA in Table 4.4. These
price assumptions show a steady growth. The answer is sensible. As
excess capacity continues to be very low, price needs to ration the
markets demand. To get 109.50 million barrels of oil out of the
ground in 2035, many new fields, unprofitable at todays prices,
would require the ability to contribute to the global liquids produc-
tion mix. Before the great recession that collapsed the markets in
2008, price raced towards US$147/bbl, an incredible feat for a
commodity that hit a low of around US$17/bbl in 2002. There was
the push of Chinese demand, the faster decline in Mexican, North
Sea and US production, and a dwindling of excess capacity to a point
where only 800,000 bbls/day was projected to stand between easily
functioning markets and an aggregate stock-out.
107
Table 4.3 International liquids supply and disposition summary (million barrels per
day)
2009 2010 2015 2020 2025 2030 2035 Annual

growth (%)
Liquids consumption 201035
OECD
US 50 states 18.81 19.17 19.1 19.02 19.2 19.47 19.9 0.10
US territories 0.27 0.28 0.31 0.32 0.34 0.36 0.36 1.00
Canada 2.16 2.21 2.15 2.21 2.25 2.29 2.35 0.20
Mexico and Chile 2.35 2.34 2.39 2.43 2.5 2.6 2.68 0.50
OECD Europe 14.66 14.58 14.14 14.43 14.65 14.76 14.74 0.00
Japan 4.39 4.45 4.51 4.6 4.62 4.51 4.42 0.00
South Korea 2.15 2.24 2.25 2.35 2.46 2.53 2.56 0.50
Australia and NZ 1.16 1.13 1.11 1.14 1.17 1.21 1.23 0.20
TOTAL OECD 45.94 46.4 45.95 46.5 47.19 47.72 48.24 0.20
NON-OECD
Russia 2.73 2.93 3.02 2.94 2.91 2.94 2.97 0.10
Other Europe and
Eurasia 2.15 2.08 2.3 2.35 2.45 2.55 2.63 0.90
China 8.33 9.19 12.1 14.36 16.03 17.65 18.5 2.80
India 3.11 3.18 3.7 4.58 5.4 5.79 5.8 2.40
Other non-OECD Asia 6.43 6.73 7.28 7.95 8.85 9.4 9.89 1.50
Middle East 6.84 7.35 7.78 7.69 8.16 8.98 9.49 1.00
Africa 3.23 3.34 3.3 3.37 3.57 3.8 4.09 0.80
Brazil 2.52 2.65 2.84 2.94 3.15 3.47 3.8 1.50
Other Central and
South America 3.07 3.19 3.49 3.66 3.81 4.05 4.09 1.70
Total non-OECD
consumption 38.41 40.65 45.82 49.83 54.32 58.62 61.26 1.70
Total liquids
consumption 84.35 87.05 91.76 96.33 101.51106.35 109.5 0.90
OPEC Production 33.34 34.58 37.3 39.23 41.91 44.05 45.89 1.10
Non-OPEC production 51.01 52.47 54.46 57.1 59.6 62.3 63.61 0.80
New Eurasia exports 10.25 10.53 11.11 12.6 13.94 14.85 15.54 1.60
OPEC market share
(percent) 39.5 39.7 40.7 40.7 41.3 41.4 41.9
Source: EIA, Annual Energy Outlook 2012, Table A21
The demand destruction that ensued from the recession

temporarily changed the equation; however, does this risk still exist?
Just as Hubbert predicted, in early 2008 an almost universal feeling
of peak oil and high prices were beginning to be the norm. Then
108
Table 4.4 Production
2009 2010 2015 2020 2025 2030 2035 Growth

(%)
Crude prices (2010 US$/BBL)

Low sulphur light 62.37 79.39 116.91 126.68 132.56 138.49 144.98 2.40
Imported 59.72 75.87 113.97 115.74 121.21 126.51 132.95 2.30
Crude oil prices (NOM)
Low sulphur light 61.65 79.39 125.97 148.87 170.09 197.1 229.55 4.30
Imported 59.04 75.87 122.81 136.02 155.52 180.06 210.51 4.20
Petroleum liquids production

OPEC
Middle East 22.3 23.43 25.46 27.16 29.77 32.07 33.94 1.50
North Africa 3.92 3.89 3.62 3.42 3.37 3.31 3.27 0.70
West Africa 4.16 4.45 5.09 5.35 5.4 5.31 5.26 0.70
South America 2.43 2.29 2.13 1.97 1.92 1.79 1.72 1.10
Total OPEC prod 32.8 34.05 36.3 37.91 40.46 42.48 44.19 1.00
Non-OPEC
OECD
US 8.27 8.79 9.82 10.73 10.53 10.57 10.15 0.60
Canada 1.96 1.91 1.79 1.82 1.82 1.81 1.78 0.30
Mexico and Chile 3 2.98 2.65 1.97 1.58 1.65 1.68 2.30
OECD EUROPE 4.7 4.36 3.7 3.33 3.15 3 2.83 1.70
Japan 0.13 0.13 0.14 0.15 0.15 0.15 0.16 0.70
Aust and NZ 0.65 0.62 0.55 0.54 0.54 0.53 0.53 0.60
TOT OECD PROD 18.71 18.8 18.65 18.54 17.78 17.72 17.14 0.40
Non-OECD
Russia 9.93 10.14 10.04 10.54 11.06 11.62 12.16 0.70
Other EUR AND EURASIA 3.12 3.22 3.67 4.01 4.37 4.52 4.54 1.40
China 3.99 4.27 4.29 4.46 4.79 4.93 4.7 0.40
Other Asia 3.67 3.77 3.79 3.55 3.38 3.17 3 0.90
Middle East 1.56 1.58 1.43 1.31 1.18 1.06 0.97 1.90
Africa 2.44 2.41 2.4 2.54 2.68 2.7 2.68 0.40
Brazil 2.08 2.19 2.72 3.34 3.87 4.21 4.45 2.90
Other Central and South American 1.9 2.01 2.29 2.32 2.47 2.67 2.65 1.10
Total non-OECD prod 28.69 29.59 30.63 32.07 33.8 34.88 35.15 0.70
Total liquids prod 80.21 82.44 85.58 88.52 92.04 95.08 96.47 0.60
Other liquids prod

US 0.75 0.9 1.05 1.34 1.62 2.08 2.59 4.30
Other North American 1.69 1.93 2.51 3.08 3.75 4.46 5.16 4.00
OECD EUROPE 0.22 0.22 0.23 0.24 0.26 0.27 0.28 1.00
Middle East 0.01 0.01 0.17 0.21 0.24 0.24 0.24 14.50
Africa 0.21 0.21 0.28 0.37 0.38 0.39 0.4 2.60
Central and South American 1.14 1.2 1.78 2.31 2.61 2.9 3.17 3.90
Other 0.12 0.13 0.16 0.28 0.61 0.92 1.18 9.10
Total other liquids prod 4.14 4.61 6.18 7.82 9.47 11.27 13.02 4.20
Total production 84.35 87.05 91.76 96.33 101.51 106.34 109.5 0.90
Source: EIA, Annual Energy Outlook 2012, Table A20
109
came the recession, and one can see in the consumption numbers in
Table 4.3 that very little (if any) growth is expected between 2008 and
2015. The shale revolution coming from the US and southern Canada
then appeared. At US$50/bbl, these technologies are not financially
viable, but, at US$7080/bbl, they are profitable. Once again, the
peak oil whispers have faded because of technology and may stay
quiet for a while if this technology becomes a universally accepted
means of production. However, our new pricing regime is in place.
The price assumptions made by the EIA exist so that the market stays
balanced. This theme is an important one. As we move from one
price regime to another, the effects of the market pricing is to ration
demand (as it has already done in many OECD countries since 2008)
and to price in new technologies for production that become finan-
cially viable at higher price points.
CONCLUSION
In summary, the global landscape of the market for crude oil has
many intricate influences, stemming from grade, location, politics
and its reception from its downstream counterparts at the refinery
level. The growth in emerging economies have shaken the stability of
the existing supply/demand balances, but have also ushered in a
new era boasting new methods of combating the continuous struggle
for the globe to be well supplied with crude oil. However, even as
Hubbert had predicted back in 1956, the decline of crude oil as our
main source of energy has been wildly overestimated. The cost and
the technological breakthroughs continue to preserve this
commodity as a large part of our daily lives.
1 International Energy Agency, 2012, Oil Market Report, November.

2 US Energy Information Administration, 2012, World Oil Transit Chokepoints, August 22.
3 Platts, 2011, Dated Brent: The Pricing Benchmark for AsiaPacific Sweet Crude Oil, May.
REFERENCES
Carlson, M., Z. Khokher and S. Titman, 2007, Equilibrium Exhaustible Resource Price
Dynamics, Journal of Finance, American Finance Association.
Evans, L. and G. Guthrie, 2009, How Options Provided by Storage Affect Electricity
Prices, Southern Economic Journal, 75(3), January, pp 681702.
110
Hubbert, M. King, 1956, Nuclear Energy and the Fossil Fuels, Shell Development
Company, Publication Number 95, presented before the Spring Meeting of the Southern
District, American Petroleum Institute, San Antonio, Texas, March.
Kaldor, N., 1939, Speculation and Economic Stability, The Review of Economic Studies.
Oliver, M., C. Mason and D. Finnoff, 2012, Pipeline Congestion and Natural Gas Basis
Differentials: Theory and Evidence, University of Wyoming.
Samuelson, P., 1965, Proof that Properly Anticipated Prices Fluctuate Randomly
Industrial Management Review, 6.
Working, H., 1948, Theory of the Inverse Carrying Charge in Futures Markets, Journal of
Farm Economics, 30, pp 128.
Working, H. 1949, The Theory of Price of Storage, American Economic Review, 39, pp
1,25462.
111
5
Wholesale Power Markets
William Webster
RWE Supply and Trading
The objective of this chapter is to provide an understanding of how

the wholesale electricity market functions, and to explain its special
features compared to other commodity markets. Despite the liber-
alised electricity markets having their first beginnings as far back as
the 1990s, there probably remain few people outside of the industry
who conceive of electricity as a traded commodity. This can be easily
discerned from political discussions where there is regular pressure
on government and regulators to intervene in the setting of elec-
tricity prices.
However, an unhindered liquid wholesale market that sets prices
is an essential component of a competitive market for electricity.
Otherwise new suppliers and new generators cannot enter the
market independently. This means all the usual components of
commodity markets need to apply: the free interaction of supply and
demand, development of forward markets, the participation of a
diverse range of traders with different motivations and strategies,
and the provision of platforms offering a range of matching, clearing
and settlement services. This chapter will describe how these basic
building blocks of traded commodity markets are applied in the elec-
tricity sector, and examine some of the outcomes.
The following section will explore some of the special features of
electricity and how they have influenced the development of whole-
sale markets, before we look at how electricity is traded in practice
and introduce some of the products and markets that are typically
found. We will then examine the behaviour of different market
participants and explore some trading strategies, as well as review
113
the development of market prices over time in some important

European markets. The chapter will also seek to identify some key
issues that might affect electricity trading over the next decade, and
end by considering some of the main sources of information on elec-
tricity wholesale markets.
ELECTRICITY AS A COMMODITY
The unique characteristics of electricity
The scientific laws of electricity
The power market has particular characteristics that distinguish it
from other commodity markets. These characteristics are mainly a
consequence of the scientific laws of electricity production, transmis-
sion and consumption.
These laws mean that, for example, it is not straightforward to
trace the production and use of individual electrons across the trans-
mission and distribution networks. Likewise, these laws mean that
the whole system has to be maintained at a constant frequency for
power plants and appliances to continue to function. There is there-
fore an interdependency between market participants that is not
seen in other sectors.
However, as with the peculiarities of other commodities, it is
possible to develop a traded market by introducing some approxi-
mation around the consequences of these physical laws. Just as the
market for crude oil is able to deal with, for example, different
quality grades and delivery locations, so it is also possible to get
around the specificities about electricity as a product. So, although
the electricity system as a whole has to balance on a second-by-
second basis, traded markets usually allow for market participants to
balance over a 15- or 30-minute period. These issues will be
discussed in more detail in the remainder of this section.
Dispatch arrangements
First, compared to other commodities, delivery of electricity is
strongly time dependent. It must be produced and delivered exactly
as it is used. This contrasts with other commodities that can be stored
to a greater or lesser extent. Electricity is also unlike most other
commodities in that it has a dedicated delivery network: the trans-
mission system. For electricity provision as a whole to continue to
function, there must be equilibrium between the network, produc-
tion and consumption in real time.
114
WHOLESALE POWER MARKETS
Second, if there is a failure in the overall system, it will affect a

broad range of users, and not necessarily those that caused the
failure. There is therefore a strong public-good element in electricity
supply. In particular, the electricity network can be characterised as
non-exclusive. If the system as a whole works, it is there for every-
body and nobody can be excluded from using it. However, electricity
supply is not a pure public good, in that it is not non-rival (in the
same way as, for example, street lighting). It is therefore competitive
with respect to supply and consumption in that the same MWh
cannot be used twice. This means that a market structure can func-
tion in the sense that the use of electricity can be rationed through the
price mechanism.
The main issue raised by these two points is, therefore, more about
the extent to which producers and consumers can interact directly, as
in other commodity markets, or whether there needs to be a specified
regulated intermediary.
In some jurisdictions, regulators impose a strong role for the trans-
mission system operator (TSO) in overseeing the market process,
and even in operational decisions. Under such arrangements, gener-
ators feed in all their technical and pricing information to the TSO,
who then calculates prices using this information and assumptions
about demand. Such market arrangements are characterised as
central-dispatch because the system operator decides how all
generation plant is dispatched on the basis of the prices and technical
information that is submitted. In effect, the TSO buys electricity on
behalf of retail suppliers and their consumers.
Meanwhile, market arrangements where producers and
consumers (or usually their retail suppliers) interact independently
are termed self-dispatch. In these cases, generators negotiate indi-
vidually with retail suppliers via traditional traded wholesale
markets structures ie, a variety of traded platforms and exchanges
as well as voice-broking services. The system operator then takes a
residual role in that they may adjust generation output via balancing
actions and re-dispatch if this is necessary to ensure the overall
security of the system.
A simplified summary of these terms is provided below.
Central dispatch
Generators provide price and technical information (eg, ramping
parameters, start costs) to the system operator. The system
115
operator compiles an efficient dispatch schedule on the basis of

this information and expected demand. Generators run to that
schedule. The TSO calculates a price for each (eg, hour) and all
trading is based around that price (eg, Ireland, England & Wales
Pool).
Self-dispatch
Retailers contract in the market with producers to meet the needs
of their portfolio of customers. Generators offer prices to the
market based on their plant characteristics and conclude transac-
tions on a bilateral basis or through an anonymous exchange.
Trading is continuous and dispatch decisions can be continuously
updated until a gate closure specified by the TSO. At gate
closure, a final dispatch schedule is notified by the generator to the
transmission system operator.
Balancing actions and re-dispatch
If, on the basis of the aggregate of final notifications, the system is
out of balance or internal security limits are breached, the system
operator will require some generators to change their actual
output from the final notification amounts. This is usually based
on priced offers by generators to increase/decrease production
compared to notified amounts.
Locational issues
The production and consumption of electricity also has a locational
element. However, it could be argued that this aspect is less impor-
tant for electricity than for other commodities. Depending on the
characteristics of the transmission network, it is not always necessary
to deliver electricity exactly to the point of consumption. Provided
the network is meshed enough, it is normal for most trading to be
conducted around particular hubs, or on a zonal basis.
With a zonal market, common in Europe, the assumption is that

transmission capacity is always available to deliver the energy to
the customer, wherever it is in that zone. This often requires
remedial actions by system operators, such as re-dispatch
(discussed above). But, as long as these do not become too
frequent or costly, these actions can take place outside the
market without upsetting trading.
The main alternative, used in North America, is a nodal market
where each node in the transmission network has a separate
116
individual price. A hub price may then develop based around a

set of nodes that usually end up with the same price, and which
are then treated as a price zone by market participants. In this
model, market participants carry the locational basis risk of
possible price changes between nodes. However, system opera-
tors also sell transmission rights between these nodes to help the
market manage these risks.
Electricity quality
Unlike many other products, electricity has the same quality for
each unit of production. One megawatt hour (MWh) is exactly the
same as another unlike, for example, natural gas where a cubic
metre of gas might have a different calorific content. However, this
physical reality has latterly been changed by environmental consid-
erations. Consumers and governments may now place a higher
value on units that are renewable or low carbon. This is already
starting to make the trading of electricity more complicated. For
example, under so-called green certificate schemes, retail suppliers
have to purchase such certificates alongside the electricity they need
in order to serve final consumers. Likewise, under other support
schemes, renewable energy might be sold in wholesale markets on a
must-run basis, even if prices are zero or even negative. The fact
that a section of the electricity market is asked to behave in a non-
commercial manner makes it more difficult to form expectations
about spot prices and discourages forward trading.
Electricity market design

Overall, electricity markets are probably more complicated than
other commodity markets. This often raises the question about
whether they are, in fact, too complicated to allow for a normal stan-
dardised and commoditised set of products to develop. Electricity
markets are already not particularly liquid compared to other
commodities. If the market becomes further fragmented into
different time, location and quality characteristics, the future for
standardised trading begins to look rather uncertain.
In the meantime, these features normally mean that the wholesale
market for electricity is, to an extent, something of an abstract regula-
tory construction. Academic and regulatory literature often speaks
of market design for electricity, which is not a term commonly
117
used for other commodity markets. Nobody ever talks about crude
oil market design in a regulatory sense. Complications such as
freight costs and quality standards are left up to the market partici-
pants to sort out for themselves.
Part of the challenge in electricity market design is getting the
balance right between the role of the market and that of government
and regulators. Policymakers continue to struggle with this chal-
lenge, even in the most mature electricity markets. Indeed, there is an
observable cycle backwards and forward between more regulated
and more market-based policy frameworks.
Where can functioning wholesale markets be found?

At the time of writing, there are several functioning and reasonably
liquid wholesale markets that perform the central tasks of price
discovery, offer hedging opportunities and give signals to market
participants for efficient operational and investment decisions.
Liquid wholesale power markets exist to a greater or lesser extent in
several areas of the European Union, in parts of North and South
America, and in Australia and New Zealand. Traded electricity
markets are also coming into existence in other countries. This
chapter will concentrate on the development of wholesale power
markets in Europe, particularly in Germany and Britain (GB).
HOW POWER IS TRADED THE CHARACTERISTICS OF

EUROPEAN ENERGY MARKETS
European market design principles: The importance of the
balancing regime
European market design is based on self-dispatch rather than
centralised dispatch of power production unlike, for example, most
North American markets. It is therefore a bilateral two-sided market
in that generators sell into, and retailers buy from, wholesale markets.
As discussed, this means that the system operators role is
restricted to dealing with residual imbalances in the system as a
whole and resolving any locational constraints. This takes place after
gate closure, which is normally one hour before real time opera-
tion. However, in reality, system operators sometimes have to begin
to take some action before gate closure if plant expected to be used
for balancing or re-dispatch needs to be ramped up or warmed in
advance.
118
Market participants on both the generation and the retail side have
to balance at gate closure across a so-called settlement period of
either 30 minutes or 15 minutes. Those market participants whose
actual measured injections do not match their consumption are said
to be out of balance and are subject to imbalance charges. They
have to pay the system operator for the actions required to balance
the system. This payment is governed by the national regulator in the
country concerned. It is usually based on the costs to the TSO of
resolving imbalances, although the formula used varies in each
country. Balancing arrangements are increasingly market-based,
with the settlement price based on bids and offers from those gener-
ators with spare capacity, or alternatively demand-side offers.
An important consequence of this market design is that trading of
electricity and also price formation is strongly driven by the desire of
market participants to avoid the consequences of being out of
balance. If a company goes into gate closure with a short position,
they are potentially exposed to very high imbalance prices at partic-
ular times. Likewise, being long at gate closure is not without risks
either, particularly if imbalance prices can go negative, which is a
possible outcome. The balancing mechanism is therefore at the heart
of European market design.
Day ahead and intraday markets

The other main reference price in European markets comes from the
day-ahead markets. These are largely two-sided cleared auctions
operated by dedicated market operators. For example, in Germany
and France the auction is run by EPEX Spot (a joint venture between
EEX and Powernext). Meanwhile, day-ahead auctions in GB and in
Nordic markets are operated by Nord Pool Spot. The Dutch day-
ahead auction is operated by APX-ENDEX (now a subsidiary of
ICE), who also operates a day-ahead market in GB.
Day-ahead exchanges are not usually compulsory marketplaces.
However, there is a strong regulatory push to ensure these markets
are liquid. In the draft European network code on capacity allocation
and congestion management (CACM), it is envisaged that these day-
ahead exchanges will play a central role in allocation of cross-border
transmission capacity. This process is known as market coupling.
The CACM network code was slated to become binding European
legislation in 2014.
119
As well as the day-ahead markets, there are various platforms for

intraday trading. Unlike day-ahead, which is almost exclusively an
exchange-based auction, trading in the intraday can either be
exchange-based or a bilateral over-the-counter (OTC) market. This
often depends on the historical development of markets and regula-
tory attitudes. For example, in the Nordic countries intraday trading
is exclusively via the Elbas platform, which is provided by Nord
Pool, whereas the system used in Germany is a platform that allows
both exchange-based trading and bilateral exchanges.
Forward markets
Physical versus financial
However, the day-ahead and intraday phases are only for fine-
tuning positions. The vast majority of electricity is traded long before
this point on a wide range of forward markets of different types.
Forward products may be either physical or financial. Financial
trading are contracts for difference that are based around a day-
ahead reference price. With financial trading, a strike price is agreed
(eg, 40/MWh). If the day-ahead price is above this for example,
45 then the buy-side counterparty will buy their power in the day-
ahead market and the seller of the forward product will pay them the
5 difference. The buyer does not take on any obligations with
respect to balancing and nomination, as discussed earlier.
Physical contracts are used when both parties are already respon-
sible for balance. Then the transaction is an obligation on the selling
party to physically deliver the amount sold or else face the imbalance
charges on behalf of the buyer.
Brokers such as Trayport and Spectron offer a screen-based broker
service based on physical delivery. Other products, such as those
offered by EEX, APX-ENDEX or Nasdaq, are financial trades based
on contracts based on the day-ahead prices. Voice-activated trading
is also possible.
Forward market products

There is a range of possible products for forward trading. The most
liquid market is for baseload power, meaning a flat amount of power
over a 24-hr period. Baseload power can be traded weekly, monthly,
quarterly, by seasons or annually. Trading may be either exchange-
based and cleared, or through bilateral OTC transactions. Trading in
120
seasonal and annual baseload products usually goes out to 23 years

into the future for both financial and physical settlement.
The other main product traded forward is peakload. This refers to
the period of 07002300 each day. Again there is a range of forward
peakload products available. However, the forward curve is not as
liquid as for baseload. Products are usually only available for 12
years in advance of real time.
Finally, it is also possible to trade four-hour blocks in some
European markets, such as in the GB market. However, these are
usually not available until some days/weeks before real time.
Spark and dark spreads

The final complication to mention is that trading in baseload prod-
ucts, in particular, is largely on the basis of spreads. For example,
the spark spread is the difference between the electricity price and
the cost of producing that electricity from a certain standard effi-
ciency gas-fired power plant, based on the prevailing gas prices. The
dark spread is the same concept for coal. With the advent of carbon
trading, indexes for clean spark spread and clean dark spread
were developed which are popular forward products, particularly in
the GB market where both coal and gas have liquid reference prices.
HEDGING STRATEGIES AND PRICE FORMATION

Market participants will usually have some pre-specified procedures
about how they interact with wholesale markets. This will partly be
driven by the companys risk controls. No company will wish to take
or maintain a position that will leave it too exposed to a disadvanta-
geous movement in prices. In particular, taking on large exposed
positions requires the company to allocate risk capital to trading
activity that is earmarked to cover possible adverse price move-
ments. In addition, accounting rules, specifically the International
Financial Reporting Standards (IFRS), may also discourage compa-
nies from taking large positions since these have to be marked to
market in a companys account. This can result in a potentially large
impact on the company P&L, with undesirable knock-on effects on
credit rating and market sentiment.
In general, the expectation is that retailers hedge the bulk of their
positions in advance through trading of baseload and peakload
products. They will then use the short-term markets for fine-tuning
121
their exposures. They may have some kind of target hedge path in
terms of what proportion of their consumers needs should be
covered by a certain date eg, that X% should be bought by Y
months before consumption.
Likewise, generators will also sell the bulk of their generation
capabilities in forward markets in order to allow for effective busi-
ness management. For example, the generation business will need to
know in advance how much revenue they are likely to collect in a
particular year. They will then be able to decide on a maintenance
timetable and other budgeting decisions. However, they will not
necessarily sell all potential volumes into forward markets since this
implies a risk in the event of a generation failure.
In essence, price formation in forward markets, and therefore
customers bills, is the consequence of how these decisions are taken
about how, and when, to buy and sell. For example, the more that the
supplydemand position is expected to be tight, the more that
retailers will tend to try and manage their exposure to short-term
markets and seek to hedge earlier, pushing up forward prices.
Conversely, if there is expected to be large margins of spare genera-
tion capacity, retailers may be more content to delay buying volumes
and wait for prices to fall. Similarly, generators may have to accept
selling at lower spreads if they see a lot of spare generation capacity
around and there is little prospect of prices increasing in spot markets.
HISTORICAL PRICE PERSPECTIVE

Germany
Figures 5.1 and 5.2 show the main trends in electricity prices in
Germany. The German electricity market is the most liquid in
Europe, if not the world. Trading is based on a single Germany/
Austria day-ahead reference price.
Initially, market opening between 2000 and 2005 led to significant
reductions in wholesale market prices as more competition was
introduced and trading became established. Prices gradually
increased between 2005 and 2008, bringing considerable new invest-
ment in generation. Some 10GW of new conventional plant began
operation in the period 201013. However, the financial crisis and
reductions in industrial demand have bought about significant price
reductions. This was only partly reversed by the enforced closure of
all German nuclear plants in 2011 after the Fukushima disaster.
122
Figure 5.1 Germany year-ahead forward prices (200513)

140
Baseload Peakload
120
100
EUR/MWh
80
60
40
20
0
05 6 7 8 9 10 11 12 3
20 00 00 00 00 20 20 0 01
/ /2 1/
2 /2 /2 1/ / /2 /2
/0
1 01 /0 /0
1 01 /0 01 01 01
03 03
/
03 03 03
/
03 0 3/ 0 3/ 03
/
Source: RWE internal data
Figure 5.2 Germany year-ahead baseload forward clean spreads (200513)

40
30
20
EUR/MWh
10
0
6
07
09
10
12
3
5
1
01
00
01
00
20
20
20
20
2
/2
/2
/2
/2
1/
1/
1/
1/
1/
1
01
1
1
-10
/0
/0
/0
/0
/0
/0
/0
/0
01
01
01
01
01
01
01
01
01
-20
-30
Baseload clean dark spread Baseload clean spark spread
The other important feature in the German market is the

combined impact of increased renewable production and energy
efficiency initiatives. From 2010, renewable production started to
have a profound impact on the market mainly due to the sheer scale
123
of investment in this sector. Take-up of renewables has been rapid as

producers benefit from a guaranteed feed-in-tariff. Compared to
peak consumption of around 80GW, there is now some 30GW of
wind production. Meanwhile, solar photovoltaics capacity increased
from 10GW in 2010 to 30GW in 2012. In Germany, renewable
producers do not themselves sell their own production. Neither do
they have to balance their portfolios like other market participants.
Instead, the TSOs have to accommodate all renewable production,
which they themselves sell on day-ahead and intraday markets. This
is known as priority dispatch. The high installed capacity of
renewables means there are now frequent incidences where most, or
all, of electricity consumption is served by renewable production.
Understandably, this affects price formation on both spot and
forward markets. Spark spreads have become particularly weak and
have been negative since the start of 2012. The impact has been
particularly strong on peakload prices, with the difference between
baseload and peakload prices narrowing. This is because normal
peak periods have been offset by high levels of solar production
during the afternoon period in some parts of the year. In general, as
renewable penetration continues to increase, the classic baseload and
peakload products may begin to lose their relevance and alternative
products may need to emerge in order for the market to fulfil its
functions effectively.
To an extent, periods with high renewable production can be
offset by imports and exports of power to neighbouring countries.
Since 2009, Germany has participated in the centralwestern Europe
(CWE) Market Coupling project. This uses the day-ahead power
exchanges to allocate cross-border capacity such that power auto-
matically flows from low prices areas to higher priced areas. This
may help the transition of markets to the high renewables world.
Great Britain
Figures 5.3 and 5.4 illustrate similar data for the GB market. As for
Germany, there is a single price zone that covers all of the island of
Great Britain.
GB prices have followed a fairly similar pattern to those in
Germany. The fall in demand in GB was, if anything, more
pronounced than in Germany with an abrupt negative effect on clean
spark spreads. Capacity margins are such that forward prices at the
124
Figure 5.3 GB year-ahead forward power prices (200513)

120
Baseload Peakload
100
80
/MWh
60
40
20
0
05 6 7 08 09 10 11 2 3
0 00 00 0 20 0 20 01 20
1
/2 1 /2 /2 /2 1/ /2 1/ /2 /
01 /0 /01 / 01 /0 01 /0 01 01
01
/
01 0 1 01 01 0 1/ 01 01
/
0 1/
Figure 5.4 GB year-ahead forward clean spark spreads (200513)

35
Baseload Peakload
30
25
/MWh
20
15
10
0
11
05
07
08
09
13
00
01
01
20
20
20
20
20
20
2
/2
2
1/
1/
1/
1/
1/
1/
1/
1/
01
/0
/0
/0
/0
/0
/0
/0
/0
/
01
01
01
01
01
01
01
01
01
125
time of writing do not show much sign of recovery despite the antic-
ipated closure of some 1015GW of generation capacity up to around
2017.
Renewable production has not yet reached the same level of pene-
tration as in Germany and its impact will continue to grow.
However, a key difference in the GB market is that renewable
producers are, and will continue to be, responsible for selling their
own power and, other than the smallest facilities, are balance-
responsible. This may prevent the impact on prices being of the same
magnitude. The subsidies for solar production and the extent of take-
up, in particular, are markedly less generous.
Compared to total peak demand of some 60GW, there is around
12GW of renewable production, a much lower percentage than in
Germany. Only around 1GW of solar photovoltaics has so far been
installed in the GB market.
Wider relationships between European markets

European markets are becoming increasingly correlated, especially
as interconnection between EU countries increases and the existing
infrastructure is managed more efficiently via market coupling.
However, there are still major locational issues and associated basis
risk that affects them.
The main locational features of European power supply is that,
due to hydroelectricity resources, the Nordic countries usually have
a year-round surplus of generation (unless there is a very cold
winter, preceded by very dry conditions). This often leads to
comparatively low wholesale prices in the Nordic system.
Both France and Belgium have high shares of nuclear power and
these countries have traditionally had low wholesale prices.
However, the high level of peak heating demand increasingly means
that these countries now import in the winter. During 200912, the
price differential between Germany and France closed and has
reversed to an extent that in France, prices are higher than those in
Germany. Both GB and the Netherlands electricity prices are typi-
cally driven by gas prices, and a locational spread with Germany will
emerge if gas and coal prices deviate. Italy has typically had the
highest wholesale electricity prices in the EU.
Differences between these regions are maintained as a conse-
quence of constraints in the overall European transmission network.
126
Generally speaking, the construction of new transmission assets is

very slow as a consequence of local resistance to new lines being
built. The main problems are objections to the visible appearance of
new transmission lines. Transmission assets are normally
constructed on a regulated basis, although there have been a few
sub-sea merchant interconnectors, such as Britned (between GB and
the Netherlands).
At the same time, the local supplydemand balance also tends to
move rather slowly as new generation assets are added and others
close. Overall, the extent of price differences between European
regions has tended to reduce slightly over time.
New developments
Power prices are increasingly driven by regulatory interventions, in
particular the objective of European Union countries to extend
renewables and to decarbonise. As already noted, the significance of
the traditional baseload and peakload divisions of wholesale prod-
ucts is beginning to be questioned. Locational issues are also
becoming more complex as there will no longer be price areas that
have low or high prices throughout the years or seasons. Instead, the
variations will tend to be increasingly seen in short-term markets.
Another regulatory development may come from possible
changes to the price zones. At the time of writing, the EU is devel-
oping network codes that will embed the methods of market
coupling that have already been in use for some time. Part of this
discussion, however, is about whether the price zones as of 2013,
mainly based on national borders, accurately reflect the real trans-
mission constraints in the network. This raises the prospect of price
zones being split, or indeed merged, in the future. This may affect
how basis spreads between different zones develop. If the price
zones more closely matched transmission constraints then the basis
spreads between zones would probably be larger and more stable.
A final important locational issue may arise from the introduction
of flow-based market coupling. This model better takes into account
the inter-relationships between use of capacity on different inter-
connectors in the meshed European networks. For example,
suppose there are three price areas: A, B and C. In reality, the
capacity available between area BC is affected by how much elec-
tricity is flowing between A and B and between A and C. A
127
flow-based approach explicitly takes these interactions into account.

With a non-flow-based approach these relationships are not
captured and the available transmission capacity between each area
is set independently.
The flow-based model will, in all likelihood, tend to make the
envelope of interconnection capacity larger. On the other hand, it
may be more difficult for market participants to understand the price
formation process and make it more difficult to formulate a trading
strategy.
KEY ISSUES FOR THE COMING DECADE

Evolving market design
The main issues for the coming decade are mainly regulatory rather
than purely economic. In particular, the increase in renewable
production will be the main challenge for the market between 2013
and 2020. First, it creates long-term uncertainty, beyond the trading
horizon, about what level of renewables penetration will occur. In
addition, the way in which renewable production is activated and
sold into the market also brings short-term issues.
Under priority dispatch schemes, as in Germany, the renewable
power tends to be sold into day-ahead and intraday markets by the
system operator rather than being spread over forward markets.
This creates unnecessary volatility and uncertainty. There are some
moves towards removing priority dispatch rules and asking renew-
able producers to sell their own production into the wholesale
market. This is expected to introduce more commercially oriented
trading strategies that will be more predictable and stable.
Meanwhile, in the GB market things are moving in the opposite
direction. Under the proposed contract for difference (CFD) scheme,
renewable producers will be compensated for the difference between
the day-ahead price and a negotiated fixed strike price. So,
although renewable producers will be required to sell their own
power, the linkage of the CFD to the day-ahead price may again
mean that plant is not being optimised in a predictable commercial
way.
Other regulatory developments include the intention, in many EU
member states, to introduce capacity mechanisms. This is part of the
policy response to the uncertainty created around the extent of
renewables and other low-carbon penetration. However, these inter-
128
ventions will inevitably have an impact on power prices. They will

also introduce a further set of regulatory uncertainties that will make
developing a trading strategy more challenging, and this is likely to
reduce the liquidity of forward markets.
Finally, as volatility moves from the forward markets to more
short-term markets, different traded products may increase in
significance. Option products are a market-driven way to reward
capacity and flexibility. There may, therefore, be greater use of
options to allow portfolios of intermittent generation to be managed
effectively. Of course, this will only happen if renewable producers
are responsible for their own portfolios and if a voluntary option
market is not undermined by regulatory interventions. Some
designs of capacity market such as the reliability options model
used in North America are effectively a compulsory, centralised
option market.
Financial market regulation

Financial regulation is also set to have an impact on the format of
trading. The EU regulation on OTC derivatives, central counterpar-
ties and trade repositories (EMIR) came into force on August 16,
2012. It includes a requirement to centrally clear transactions once a
companys portfolio exceeds a certain threshold of 3 billion. Many
large energy trading houses may be captured by this and, if so, there
will be an increase in the amount of cleared transactions as a result.
Discussions on the exact requirements were ongoing throughout
201213 via the Draft Technical Standards. These were produced
by the European Securities and Markets Authority (ESMA) and,
following the scrutiny of the European Parliament that concluded in
February 2013, they were to be adopted by the Commission as
binding requirements via the Comitology process and phased in
over three years: 201316.
In addition, discussions were ongoing during 2013 about new
versions of the Markets in Financial Instruments Directive (MiFID).
The old directive will be replaced by MiFID2 and a regulation
(MiFIR). One possible outcome is that trading houses above a certain
size will be regulated in the same way as banks, complete with strict
capital requirements. However, there are possible exemptions that
are being discussed, including the ring-fencing of some physical
trades when determining whether companies exceed the threshold.
129
MiFID is also expected to set requirements on companies regarding

position limits and risk management techniques.
Many of the proposals put forward in the EU context are already
part of the legislation in the US, via the DoddFrank Act. Traders will
have to get used to compliance with this type of regulation.
However, all these tend to add transaction costs and potentially
reduce the liquidity of wholesale markets. Other interventions have
been regularly floated, such as the Financial Transaction Tax or
sporadic restrictions on short-selling. These could have a similar
negative impact on traded markets.
Changing consumer requirements more bespoke services?

Other more consumer-driven factors are also relevant. The spread of
small-scale renewable generation may tend to move the market
away from more centralised solutions, and in the direction of more
localised and bespoke solutions. New technologies such as electricity
storage may also be more easily developed on a small scale. This will
mean that the traditional relationships between producers and
consumers will become blurred so that they become amalgamated
into one role. So-called prosumers may become much more usual.
Again this may make a centralised traded market less important. On
the other hand, the development of alternative, innovative traded
products may still preserve the role of the classic trading function.
SOURCES OF MARKET INFORMATION

There is a wide range of sources of information on European
markets. In 2005, the European Commission established the Energy
Market Observatory, which now produces regular reports on price
developments, investment, etc. The transmission system operators
(via the European Network of Transmission System Operators,
ENTSO) also provide information on interconnector availability and
regular assessment and projections of the supplydemand position.
The Regulation on Energy Market Integrity and Transparency
(REMIT) came into force in 2011, which requires all electricity and
gas companies to publish any inside information that they hold. In
effect, this means provision of data on all planned and unplanned
outages, projected return to service dates and metered production
volumes of all power plants above a certain size.
It is expected that REMIT will be strengthened during 2013 with
130
the introduction of more specific binding guidelines on transparency

from the European Commission. This will apply to generators, trans-
mission system operators and large consumers. The likelihood is that
this will lead to a centralised platform for reporting information. At
present, companies are largely reporting inside information on their
own individual websites.
CONCLUSIONS
The main questions about power markets in Europe are well known.
Where are prices and spreads going? What will the market look like
in 2020 and beyond? Will there even be a market that we recognise?
The first question is difficult to answer. Prices and spreads are
low, and this is mostly due to an unexpected event: the financial
crisis and its impact on the economy and electricity demand. So,
although at the time of writing there does not seem much prospect of
recovery, we do not yet know what other unexpected events might
occur. However, we do know that prices for the traditional baseload
product are likely to be continually eroded by more renewable pene-
tration. Meanwhile, flexibility should become more valuable, so we
might end up in a situation where one type of traded product
continues to experience falling prices, while prices are rising in
another segment of the market.
The market in 2020 will clearly look somewhat different. More
complex and bespoke products may develop, which may or may not
have the same liquidity as the traditional ones. Trading might also
continue to move towards the short term as it becomes more and
more difficult to take a position on how things will look beyond one
or two years. This may feed through into the relationships between
the market and consumers. Supply contracts to end-users based on
long-term contracts may also become prohibitively expensive in
view of the additional risks and uncertainties.
Will the traded market exist at all? There is clearly some risk that
the panoply of regulatory interventions will drive liquidity out of
wholesale markets entirely. Contractual structures may then become
more bespoke and possibly also have a high degree of regulatory
involvement. More integrated solutions may become more popular
and this will move us away from traded outcomes.
131
6
The Metals Markets
Kamal Naqvi
Credit Suisse
In this chapter, we will examine the key determining factors for

metal price analysis: physical demand, supply and inventory. We
will then explore how these three factors combined lead to price
formation, together with a short discussion of a range of other influ-
ences such as currency, speculative and investor flows or
positioning, and inflation. Across the metals complex, it should be
apparent by the end of this chapter that the importance of these
various factors varies significantly from metal to metal. We finally
conclude with a short discussion regarding the major differences
between the three metal segments, with a summary of the individual
fundamentals and trends in these markets.
For the purposes of this chapter, we shall define the metals
markets, also known as basic materials or industrial minerals, as
mined commodities that have a recognised and liquid global paper
trading market that is widely used as the primary pricing mecha-
nism for that commodity. The metals markets, under this definition,
can be split into three areas: base metals, bulk commodities and
precious metals. However, for much of this chapter we shall refer to
the entire group as metals.
The metal markets are, arguably, the most direct expression of
applied macro and microeconomics. The core driver of demand for
almost all metals is industrial production, on a country, regional and
world basis. However, there are micro differences for individual
metals demand and these can be very important for idiosyncratic
pricing. The nature of global metal supply tends to be relatively
133
stable with, typically, only modest seasonality compared to, say,

agriculture. The meeting of demand and supply then is the stock and
flow of inventory, which is the main underpinning for metal prices.
The global metals markets are one of the longest serving
commodity trading markets. Used as a currency at various points of
history across the world, the metals markets are now best understood
as, arguably, the purest form of global commodity market due to their
homogeneity. Unlike most agriculture and energy markets, the metal
markets tend to have largely standardised physical properties and are
less specific to regions or countries. Hence, metal prices tend to reflect
the interaction between global supply, demand and inventories.
INVENTORY
As they are relatively easily stored, inventories for metals tend to be
more visible and therefore quantifiable compared to other commodi-
ties. The key to fundamentally driven commodity pricing is the
relationship between inventories and price. For most of the metals
(gold is perhaps an exception), this is a normal relationship with
declining inventories typically associated with upward price pres-
sure. This is shown in Figure 6.1.
The two key elements for pricing dynamics are:
the level of inventories, measured best in terms of how many

days, weeks, months or years of consumption; and
the rate of change in inventory levels.
These two factors combine to form the physical fundamental drivers

for metal prices. A very low level of available inventories, such as
copper or tin (as noted in Table 6.1), will typically see high and
volatile prices as in this situation only modest changes in the
supply/demand balance are needed to produce a large change in
prices. In contrast, metals with very large levels of inventories, such
as gold and silver in Table 6.1, require much larger changes in the
supply/demand balance to justify a change to price.
It should be noted that the weeks of consumption heading in
isolation means little for relative pricing, but is shown for illustration
of relative availability of metal inventories. The price for an indi-
vidual metal depends more on the relative level of inventory
compared to its own history.
134
THE METALS MARKETS
Figure 6.1 Commodity prices and inventories

120
100
Community price ($/t, $/oz, etc)
80
60
40
20
0
0 2 4 6 8 10 12 14 16
Weeks of consumption
Source: Credit Suisse, Wood Mackenzie
Table 6.1 Commodity inventories by weeks of consumption
Commodity Weeks of consumption

(2012)
Copper 1.5
Tin 2.1
Lead 3.1
Iron ore 6.0
Thermal coal 6.0
Zinc 8.0
Nickel 11.0
Aluminium 16.6
Platinum 40.0
Palladium 60.0
Silver 400.0
Gold 700.0
Inventory levels are not only the primary driver for price levels
and change, but also for forward pricing, which will be discussed
later in the chapter.
DEMAND
Metals demand is strongly linked to economic growth. However,
while the level of global GDP is a reasonable proxy for living stan-
135
dards and can be a useful broad macroeconomic variable for the

energy and agricultural markets, it is not such a useful representa-
tion for industrial materials demand, as a large share of GDP is
related to the services sector in most developed economies. Rather,
the best macroeconomic drivers of industrial materials demand are
industrial production (IP) and real fixed asset investment (FAI), as
shown in Figure 6.2.
There are, of course, micro differences between the metals markets
in terms of sensitivity to these broad macro variables, depending on
which sectors and countries dominate their use (see later in the
chapter), but they are relatively modest compared to the primary
trend.
On a national level, for industrial materials one country has
become dominant: China. As shown in the Figures 6.3 and 6.4,
Chinese demand for almost all metals has become dominant in
absolute terms and even more so as a proportion of global demand
growth. For this reason, much of the traditional analysis of demand
by country has been overwhelmed by the flows in Chinese demand,
particularly as represented by Chinese trade data.
SUPPLY
Metals supply originates from mined ore that is then processed into
standardised physical properties to allow for global sale. The various
Figure 6.2 Global industrial production growth (month-on-month trend)

1.5%
1.0%
0.5%
0.0%
-0.5%
-1.0%
-1.5%
-2.0%
-2.5%
2000 2002 2004 2006 2008 2010 2012
Source: Credit Suisse, Thompson Reuters Datastream
136
THE METALS MARKETS
Figure 6.3 China is a key driver of growth in global metal demand

35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0%
70s 80s 90s 00s 10s*
* 2010s average of first four years, with Credit Suisse 8% forecast for 2012 and 2013
Figure 6.4 China dominating copper, aluminium, steel oil markets

50% Steel Copper Aluminium Oil (rhs) 12%
45%
11%
40%
10%
35%
30% 9%
25%
8%
20%
7%
15%
10% 6%
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Source: Credit Suisse BP World Statistical Yearbook, Wood Mackenzie, World

Steel Association
traded metal products, somewhat similar to the energy complex, are

a variety of extracted and processed minerals. Iron ore and coal are
concentrated ores and require only relatively modest processing to
standardise quality. Copper, zinc, lead, tin and nickel are refined
metals from concentrate, while aluminium and the precious metals
require further elaborate processing to meet global standards.
137
Figure 6.5 World copper mine production has grown very slowly since the
1990s, but this could change in 201314
22000 Mine supply (without disruption), kt 10
Mine supply, kt
Increase, % (rhs)
20000 8
18000 6
16000 4
14000 2
12000 0
10000 -2
1995
1996
1998
1999
2000
2001
2003
2005
2006
2008
2009
2010
2011
2013
2015
1997
2002
2004
2007
2012
2014
Scrap can also be a meaningful source of annual supply for some

metals, such as lead and the precious metals.
Long lead times for new mines tend to lead to longer price cycles
for many metals. In Figure 6.5, copper mine production can be seen
to have grown only modestly from 2004 to 2012, despite a massive
increase in copper prices. This is due to the lagged response of mine
supply to price.
The cost of supply is the other supply-side factor that supports
prices. Figure 6.6 depicts the industry cost curve for aluminium in
2012, and this can be used as an indication of sustainable prices in the
medium term. However, this support level is not a stationary one as
most elements of mine supply costs such as labour, power, equip-
ment and energy are also cyclical.
Using copper as an example, Figure 6.7 illustrates the drivers of
mined supply costs and also highlights the sharp escalation in costs
in copper mine supply since 2005. In money-of-the-day terms, mine
site cash costs have doubled, largely due to steep increases in the unit
costs of labour (direct wages), service provision (essentially a form of
labour) and consumables. Energy costs have also risen, but for
copper mines these are a smaller proportion of costs than, say,
aluminium production.
138
THE METALS MARKETS
PRICES
As discussed earlier, the key to fundamentally driven commodity
pricing is the relationship between inventories and price. The actual
or estimated level of inventories, best measured in terms of
consumption, and the expected change in inventories, known as the
market balance, are the core drivers for the price level, the volatility
of prices and the shape of the forward price curve.
The metals markets tend to have long price cycles due to the long
lead times in mined supply. Figure 6.8 depicts a long-term time
Figure 6.6 Aluminium cost curve (2012)

3,500
Cash cost (C1)
3,000
Cash cost (C1)($/t)
2,500
2,000
1,500
1,000
500
0
0 20,000 40,000
Production (kt/a)
Figure 6.7 Copper mine costs of production: sharp rises in consumable and
labour costs
4000
3500
3000
Services & other
2500
Stores
2000
Fuel
1500 Electricity
1000 Labour
500
0
1990 1995 2000 2005 2010 2012
139
series for base metals showing how long the price cycle tends to be
and also, interestingly, that current prices for base metals are not
significantly high in real terms. This is despite the fact that since 2002,
the prices of all metals have risen significantly, with gold and iron
ore reaching all time highs in 2012, as shown in Figure 6.9.
The rise of electronic access to commodity markets and growth in
high-speed trading technology has, in our view, changed short-term
commodity pricing dynamics not necessarily for the better or
worse, just changed. A standard technical analysis for copper, for
instance, has become a new challenge for traditional commodity
Figure 6.8 Average real base metal prices

8 Principal component Equally-weighted metals index (logs, rhs) 7.5
6 7.0
4
12 6.5
17 years!
19 years! years
2
so far..! 6.0
24 years! 20 years!
0 23 years!
19 years! 5.5
-2
5.0
-4
-6 4.5
-8 4.0
1850 1870 1890 1910 1930 1950 1970 1990 2010
Source: Credit Suisse, IMF, Bloomberg Professional Service
Figure 6.9 Gold, oil, iron ore and copper remain expensive relative to history
250%
200%
150%
100%
50%
0%
-50%
-100%
Aluminium Wheat Corn Zinc T. Coal Nickel Tin Lead Copper Iron Ore Brent Gold
Crude

Note: Indexed to 2002 prices
140
THE METALS MARKETS
market participants, including (perhaps even especially) specialist

commodity hedge funds. A paper on this topic (Filimonov, V., D.
Bicchetti, N. Maystre and D. Sornette, 2013, Quantification of the
High Level of Endogeneity and of Structural Regime Shifts in
Commodity Markets, SSRN) concluded that there is evidence of
greater price endogeneity rather than external news/factors. Other
markets have gone through the same evolution, and the metals
market is no different. It does not mean the physical commodity
fundamentals have become irrelevant indeed, return dispersion
suggests the opposite it simply means that there are a few more
variables added to the market.
Macro factors have latterly become a more important driver of, or
rationalisation for, metal prices. The two factors that have endured
the cycles as being an influence on metals prices, or being influenced
by metal prices, are currencies and inflation.
Figure 6.10 shows a long-run series of copper prices in a variety of
currencies; it is notable that for key cycles the price experience can
diverge significantly. This is relevant to metal price formation, as a
weak domestic currency is a positive for producers and a negative
for consumers, with the oppositive also being the case.
The link between metal prices and inflation is more muted for
most metals, with the clear exception for gold. For many reasons,
gold is an exception to the price formation basis for the majority of
the metals markets. It has often been seen as a long-term preserver of
Figure 6.10 Currency appreciation significantly affected copper prices

700
USD AUD JPY
600
500
400
300
200
100
0
1971 1981 1991 2001 2011
141
wealth and, therefore, a hedge against inflation. Certainly it is true

that, at times, gold prices can be highly correlated with inflation
expectations (see Figures 6.11 and 6.12).
The bulk of this chapter has discussed spot or front price forma-
tion, which is the prime focus for metals market analysis as it
determines the demand for physical metal for immediate delivery
Figure 6.11 Gold versus five-year TIPS (since 2007)

$2,000 -3.0
Gold, $/oz (LHS)
$1,750 US 5 year TIPS, % -2.0

(scale inverted)
$1,500 -1.0
$1,250 0.0
%
$1,000 1.0
$750 2.0
$500 3.0
$250 4.0
Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13
Figure 6.12 Gold versus five-year TIPS

$2,000 -2.0
$1,750 -1.0
$1,500
0.0
$1,250
1.0
$1,000
2.0
$750
Gold, $/oz (LHS)
$500 3.0
US 5 year TIPS, % (scale inverted)
$250 4.0
Jan-09 Jan-10 Jan-11 Jan-12 Jan-13
142
THE METALS MARKETS
from which all market price points are determined. However, when
trading the metals markets, much discussion revolves around the
point of the forward price curve that needs to be traded and to what
degree that point does or does not reflect the expectations already
priced-in to the market.
For example, in Figure 6.13, the iron ore market curve changed
significantly in both shape and level over several months reflecting
how changes in market expectations, the demand from physical
consumers for immediate delivery of metal and the flow of business
across the various points of the curve can shift and reshape the
forward prices.
Commodities with relatively low levels of available inventory
tend to be in backwardation, with nearer-dated futures contracts at
higher prices than the futures contracts further out the curve,
reflecting the premium that the consumer is willing to pay to secure
metal. If the contrary is true, and the market is perceived to be in
ample or over-supply, then the futures curve tends to be upward
sloping, and the market is said to be in contango.
Metals tend to have a somewhat more consistent contango
compared to energy due to the relative ease of storing metals. Gold is
the extreme example of this, with storage of gold being a tiny fraction
of its cost and, therefore, gold tends to trade in perpetual contango
Figure 6.13 Iron ore market curve
Iron ore 62% China (TSI) swaps : NYM : last price : 6/4/2013
Iron ore 62% China (TSI) swaps : NYM : last price : 12/5/2012
Iron ore 62% China (TSI) swaps : NYM : last price : 5/3/2013 125
USD/metric tonne
120
115
110
105
Dec 2012
Feb 2013
May 2013
Jul 2013
Oct 2013
Dec 2013
Mar 2014
May 2014
Aug 2014
Oct 2014
Dec 2014
Source: Credit Suisse, Bloomberg
143
with forward prices driven by the US interest rates minus the storage
or leasing rate. For other metals, such as aluminium, there is also a
tendency towards contango as inventory tends to be built and held
for large consumers, such as car manufacturers. Operators with their
own storage facilities and/or access to cheaper finance can some-
times buy and hold physical metal against an offsetting paper
position for a (largely) risk-free return.
BASE METALS
The base metals, also known as industrial metals or non-ferrous
metals, are aluminium, copper, zinc, lead, nickel and tin. The worlds
benchmark contracts are listed on the London Metal Exchange
(LME). However, other key contacts include the Comex Copper and
Shanghai Futures Exchange (SHFE) copper contracts.
The LME has an idiosyncratic trading system. The most active
daily price is known as the three months price, literally a trading
Figure 6.14 Structure of LME futures

Daily prompt dates Weekly prompt dates
Cash 3 months
Monthly prompts to 12, 15, 27, 63 or 123 months
6 months 12 15 27 63 123
LME Mini Tin Aluminium Lead, Aluminium

PP, LLDPE (alloy) & nickel (Primary
& Steel NASAAC & zinc copper)
Source: London Metal Exchange
144
THE METALS MARKETS
date which is three months forward of the current day, subject to it

being an official trading day (ie, not a UK holiday). Future points on
the forward curve are then traded as a spread to the three months
price. Figure 6.14 shows the structure of LME futures dates, daily out
to three months, weekly to six months and monthly out to 10 years
for some products.
The LME also remains one of the few remaining open outcry
trading markets where the official daily prices are set by the clearing
price found across the floor, as it is known. Commercial players
(mining companies, industrial users, physical merchants, end-
consumers), banks, brokers, hedge funds, and institutional investors
are all active participants.
Most of the discussion in the chapter so far applies to the base
metals markets in terms of market analysis and price formation. We
shall now contrast the bulk commodities and precious metals
markets.
We provide a chart and table summary of the main features of the
base metals markets in Figures 6.156.18 found at the end of this
chapter.
BULK COMMODITIES
For the purpose of this chapter, we limit our definition of the bulk
commodities to the mined materials of iron ore and thermal coal
(note that others may include steel and freight within the definition).
The bulk commodities are so-called due to the sheer physical
volume of production. Both iron ore and coal production are more
than the combined output of the six LME metals. However, unlike
these metals, the majority of global production of both iron ore and
thermal coal is used domestically, with the balance often being
shipped long distances to consumers. Both materials have a domi-
nant usage, with iron ore being the key ingredient for steel and
thermal coal for energy.
Historically, both iron ore and thermal coal were supplied on a
contractual basis (typically, annually), based on periodic negotia-
tions between producers and consumers. However, since the early
2000s, both markets have moved away from this structure towards a
physical spot market supported by an over-the-counter (OTC) paper
forward market. Latterly, clearing of OTC swaps and even futures
exchange markets have emerged. The OTC markets are priced
145
against industry benchmark indexes that are based upon spot phys-
ical deliveries.
Due to the magnitude of the flow and the relative high costs of
freight as a percentage of the final price, both iron ore and thermal
coal can be quoted by including the cost of freight to the consumer
port price (CFR). This is typical for iron ore, or from the port of the
producing country before the freight on board (FOB) price, which
tends to be more common for thermal coal.
PRECIOUS METALS
Precious metals, particularly gold, are among the most actively
traded commodity markets, with gold having the widest number of
trading participants of any commodity, including oil. The precious
metals that are actively traded are gold, silver, platinum and palla-
dium. All of these have liquid OTC and exchange-traded markets.
Unlike other commodities, they also have a very large physically
traded wholesale market, of which London is generally seen as the
global centre, although there is a wide range of important local
markets across the world.
The term precious relates partly to their relative scarcity and
partly as they are often used as a store of value rather than for direct
consumption although both gold and silver are commonly used as
miniature decorations on top of Indian sweets, and hence are
genuinely consumed! The precious metals markets are also distinc-
tive in having traditional banking elements that is, gold can be
deposited, on an allocated or unallocated basis, and therefore also
borrowed or leased, much like classic money.
The precious metals, and particularly gold, have probably more
trading centres than any other commodity, despite being globally
homogenous. As mentioned, while the global central point for the
precious metals market remains London, there are a wide range of
very important physical gold markets, including Zurich, Mumbai
and Dubai. However, increased market share and overall liquidity
lies in the listed exchanges, in particular the New York Mercantile
Exchange (Nymex), the Shanghai Futures Exchange (SHFE), the
Multi Commodity Exchange (MCX), the Tokyo Commodity
Exchange (TOCOM) and the Dubai Mercantile Exchange (DME).
Unlike other commodities, a large fraction of all the precious
metals mined in history still exist and can be considered, at least
146
THE METALS MARKETS
theoretically, to be above-ground inventory. This is not so much the

case in silver, and even less in platinum and palladium, which is why
they are more similar to the base metals markets.
Furthermore, precious metals, and particularly gold, used as a
central bank asset in bar form means that there is also an active and
liquid lease or borrowing market, which reduces the scope for phys-
ical scarcity to influence the price although occasionally certain
bars or coins may trade at a higher premium due to their individual
scarcity. Instead, market sentiment tends to dominate precious
metals prices and this can be influenced by many differing elements,
of which physical supply and demand is just one; others include
inflation, currencies, geopolitics and uncertainty or risk more gener-
ally. The jewellery sector is important for all the precious metals
markets, while industrial usage is also important for silver, platinum
and palladium.
Physical investor demand is also a key factor, with increased
accessibility through exchange-traded funds having become a major
market influence and now also a major inventory.
CONCLUSIONS
The chapter was primarily designed to provide an initial guide to
analysing the basic material of metal markets. It should hopefully
have become clear that while there are overarching similarities to the
group, specific analysis requires a quite idiosyncratic approach to
not only each markets supply, demand and inventory, but also to its
relationship to other commodities, particularly other metals, as well
as wider macro relationships. In reality, each individual market
could have an entire book dedicated to its analysis.
The global metals markets are at a pivotal time. Since the early
2000s, prices have often been gripped in the so-called super-cycle.
Definitions vary on what super-cycle means, but for some it is
higher than average real or nominal prices. Under such a definition,
we think this will continue. However, for most it means synchronous
rising metal prices, and this we do not think will occur. The true
super-cycle, from 2002 to 2007, was buoyed by a range of synchro-
nised, positive physical and financial factors that combined to drive
prices to historical nominal highs.
In summary, the physical factors driving the metals markets are
shown below.
147
Demand surge:
Chinese;
emerging markets; and
moderate growth across the rest of the world.
Supply constraints and costs explosion:
falling ore grades;
labour shortages and disruption;
technical problems (mines and refineries);
infrastructure bottlenecks, delays and disruptions;
resource nationalisation;
environmental and social legislation;
reduced availability of scrap; and
shift to underground mining.
Inventory declines:
falling visible exchange inventories; and
off-exchange inventories either falling or not being made
available.
Investor buying:
Investor buying:
index inflows;
structured product buying; and
exchange-traded product demand (ETFs, etc).
Hedge fund buying:
commodity specialist fund buying on constructive S&D;
macro hedge funds buying on a China play and/or US dollar
weakness; and
technical traders buying due to signals and momentum.
Corporate flows:
consumer forward buying due to concerns over price rises;
and
producer reductions of existing hedge books ie, net buying.
Looking forward, many of these factors are, or are likely to be,

much less positive; indeed, they may become negative influences on
price over the next few years. Generally, we still expect nominal and
real prices to hold in a higher range compared to history, but we also
expect to see greater variation in individual metals. The winners are
likely to be those where we see little likelihood of sustained increases
in supply such as zinc, lead, platinum, palladium and copper.
148
THE METALS MARKETS
Figure 6.15 Industrial metals: aluminium
Demand by sector
Other, 5%
Machinery &
equipment, 8% Construction, 19%
Consumer goods,
9%
Packaging, 13%
Transport, 32%
Electrical, 15%
Demand by country
50%
45% 2003 2012
45%
40%
35%
30%
27%
24%
25%
19%
20% 18% 17%
15% 14%
15%
10%
5% 3% 3% 3%
2% 2% 2% 2% 1% 1% 1%
0%
China Europe Asia North Latin Russia Middle East Africa Oceania
America America
Cost curve
3,000
90% minus 90%:

2,500
premium: US$2,072
US$1,812
US$280/t premium added 92%
US$/t
2,000 80%
Current LME
Cash: US$1,893
1,500
1,000
Source: Credit Suisse, Wood Mackenzie, International Aluminium Institute
149
Figure 6.15 Industrial metals: aluminium (cont.)

Costs breakdown
Other, 10%
Labour, 6%
Alumina, 31%
Energy, 39%
Carbon & bath, 14%
Supply by country
50%
46% 2003 2012
45%
40%
35%
30%
25%
20% 20%
20% 18%
15% 14%
10% 10%
10% 9% 8% 8%
8%
5% 5% 5%
5% 4% 4% 4% 3%
0%
China North Russia Europe Middle East Oceania Asia Latin Africa
America America
Integrated aluminium-making process flow chart

Bauxite mining
Stage 1
Alumina refining
refining
Stage 2
smelting
Recycling
Aluminium smelting
Processing Extrusion
Rolling
Casting
Source: Credit Suisse, Wood Mackenzie, International Aluminium Institute
150
THE METALS MARKETS
Figure 6.16 Industrial metals: copper
Demand by sector
Consumer products,
9%
Industrial machinery,
13% Electrical/electronics,
34%
Transportation, 14%
Construction, 30%
Demand by country
40 37.7
% of global copper demand
35
30
25
20 19.8
15
12.2
10 9.4 8.6 8.9
6.6 5.7
5.1 4.5 4.4 4.1
5 3.2 3.0 2.9
1.9 1.3 1.9 2.2 1.9
0
China USA Germany Japan South India Italy Taiwan Russia Brazil
Korea
Cost curve
10000
9000
8000
Current price = US$7,370/t
7000
6000
90th percentile = US$5,335/t
5000
4000
3000
2000
1000
0
0 5,000 10,000 15,000
Source: Credit Suisse, Wood Mackenzie, Teck
151
Figure 6.16 Industrial metals: copper (cont.)

Costs breakdown
Services/other, 23% Labour, 25%
Electricity, 13%
Stores, 32%
Fuel, 8%
Supply by country
50%
2003 2012
45% 44%
43%
40%
35%
30%
25%
20%
15%
15% 13%
9% 9% 9% 9%
10% 7% 7%
6% 6% 6%
4% 4% 5%
5%
1% 2%
0%
Latin North China Africa Russia Oceania Europe Asia Middle East
America America
CESL copper process flowchart
Evaporator Condensate
Raffinate
Oxygen
Thickener Pregnant leach solution (PLS)
Copper
concentrate
Limestone
Pressure oxidation
Wash Neutralisation
Atmospheric leach
To pressure
water (optional)
oxidation
Residue washing
(counter current Filtration
decantation)
Solvent extraction
Wash
water Filtration
Electrowinning
Leach residue Gypsum

(to gold plant) (to tailings) Copper cathode
(to market)
Source: Credit Suisse, Wood Mackenzie, Teck
152
THE METALS MARKETS
Figure 6.17 Industrial metals: nickel
Demand and industrial production

15% 16%
IP Mature economies (LHS)
IP Developing economies (LHS)
Nickel consumption (RHS)
12%
10%
8%
5%
4%
0%
0%
-5%
-4%
-10%
-8%
-15% -12%
19 6
19 7
19 8
19 9
19 0
19 1
92
19 3
19 4
19 5
19 6
19 7
19 8
20 9
00
20 1
20 2
20 3
20 4
20 5
06
20 7
20 8
20 9
10
8
8
8
8
9
9
9
9
9
9
9
9
9
0
0
0
0
0
0
0
0
19
19
20
20
Demand by country
USA
10%
Other
28%
China
33%
Germany
7%
Taiwan
5%
Korea
Japan
6%
11%
Cost curve
25,000 Ramp-ups, NPI

LME cash and tocantins
20,000 Price: US$18,917 92.7%
15,000 Median:
US$10,136
US$/tonne
10,000
5,000
0
90%:
US$15,945
-5,000
-10,000
Source: Credit Suisse, Wood Mackenzie, Nickel Institute
153
Figure 6.17 Industrial metals: nickel (cont.)

Nickel production
1400 Sulphides
Laterites
1200
1000
800
600
400
200
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
First use consumption
Others (incl. chemicals) 6%

Electroplating 11%
Other steel alloys (incl. castings) 10%
Non-ferrous alloys 12%
Stainless steels 61%
Demand by application
Tubular products 10% Other 7%
Building & construction 11%
Engineering 24%
Electro & electronic 15%
Metal goods 16%
Transportation 16%
Source: Credit Suisse, Wood Mackenzie, Nickel Institute
154
THE METALS MARKETS
Figure 6.18 Industrial metals: zinc
Demand and industrial production

10%
8%
6%
4%
2%
0%
-2%
-4%
-6%
-8%
-10%
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Global IP growth (YOY%) Est. refined zinc consumption growth (YOY%)
Demand by country
Oceania 2%
Latin America 5% Africa 1%
Japan 4%
Asia (excl Japan

& China)
18% China
41%
North America
10%
Europe
19%
Cost curve
2,500 90%:
US$1,524
US$120/tonne premium 99.1%
2,000
LME cash 95.9%
1,500 price: US$1,847
Median:
US$/tonne
US$965
1,000
500
-500
-1,000
155
Figure 6.18 Industrial metals: zinc (cont.)

Supply by country
Russian Fed.
India 2%
6%
Europe
6%
China
Other Asia 30%
7%
N. America
12%
Australia
L. America
12%
22%
First use consumption

Rolled & Miscellaneous
extruded 4%
products
7%
Oxides &
chemicals
8%
Galvanising
Decasting 57%
alloys
11%
Demand by application
Consumer
products
Infrastructure
Industrial 8%
13%
machinery
7%
Transport
23%
Construction
49%
156
THE METALS MARKETS
Figure 6.19 Bulk commodities: iron ore

210 Iron ore (62% Fe CFR Tianjin spot) Quarterly avg forecasts
190
170
150
US$/t
130
110
90
70
50
2009 2010 2011 2012 2013 2014
140
Spot Price
120
US$ per dry metric tonne
100 Consensus
CS price forecast
80
60
40
20
0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Million tonnes per annum
BHP.AX CLF FMG.AX KIOJ.J RIO.AX VALE.N China Other Reported Cash Cost (FOB)
All-In Cash Cost (FOB)
All-In 62% IODEX equiv (CFR)
Source: Credit Suisse, Wood Mackenzie, Company data
157
Figure 6.20 Bulk commodities: coal

Types of coal
CARBON/ENERGY CONTENT OF COAL HIGH
HIGH MOISTURE CONTENT OF COAL
Low rank coals Hard coal

47% 53%
% OF WORLD RESERVES
Lignite Sub-bituminous Bituminous Anthracite

17% 30% 52% -1%
Thermal Metallurgical
Steam coal Coking coal
Largely power Power generation Power generation Manufacture Domestic/

USES
generation Cement manufacture Cement manufacture of iron industrial

Industrial uses Industrial uses and steel including
smokeless fuel
Major contributors to seaborne demand
50 India RoW China
40
30
20
10
-10
2011 2012 2013 2014 2015
Major contributors to seaborne supply

40 Australia RoW Indonesia
35
30
25
20
15
10
5
0
2011 2012 2013 2014 2015
Source: Credit Suisse, Wood Mackenzie, World Coal Institute
158
THE METALS MARKETS
Figure 6.21 Precious metals: gold
Above ground stocks

60,000 Cumulative supply used as investment
Cumulative supply used in jewelry

50,000
Near to
market
Cumulative supply used in industry/dental
40,000
Annual mine supply
Tonnes
30,000
20,000
Far from
market
10,000
0
2000 2002 2004 2006 2008 2010 2012
Demand by sector
Bar coin retail

investment
26%
Dental
1% Jewellery
57%
Industrial
11%
Cost curve
1800 C3 costs (real)
1600 Average gold price (real)
1400
$/oz Au
1200
1000
800
600
400
200
1980 1985 1990 1995 2000 2005 2010
Source: Credit Suisse, Wood Mackenzie, GFMS, Thomson Reuters, World Gold
Council, Bloomberg Professional Service
159
Figure 6.21 Precious metals: gold (cont.)

Supply by sector
Old gold
scrap
39%
Mine
production
60%
Official sector
sales, 1%
Mine supply by country
Central bank reserves

34
9,000
8, 1
8,000
7,000
6,000
Tonnes
06
5,000
91
3,7
17
14
3, 3
52
35
4,000
03
2,8
2, 8
2,4
2,4
2, 3
3,000
54
40
996
1,0
1, 0
2,000
765
613
558
502
445
424
383
366
323
310
287
282
280
228
1,000
80
0
United
2012A Mine
Italy
India
Other
UK
2012A
IMF
China
Taiwan
Belgium
Switzerland
Turkey
Austria
Australia
Germany
Netherlands
Venezuela
Lebanon
Spain
Portugal
France
Saudi Arabia
Canada
ECB
Russia
Japan
160
THE METALS MARKETS
Figure 6.22 Precious metals: silver

Supply sources
Scrap
Mine 20%
production
77%
Government sales
3%
Mine supply sources

Gold 11%
Others 1%
Primary silver
28%
Zinc & lead

37%
Copper
23%
Supply trends
Mine production Net official sector sales Silver scrap
Producer hedging Implied net dis-investment Zero growth
1.0% growth rate 2.5% growth rate 5.0% growth rate
1,200 1,200
1,000 1,000
800 800
600 600
400 400
200 200
0 0
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10
11
12
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
20
Source: Credit Suisse, GFMS, Silver Institute
161
Figure 6.22 Precious metals: silver (cont.)

Demand trends
Industrial applications Photography Jewellery & silverware Coins & medals
1,000
900
800
700
600
500
400
300
200
100
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Demand by sector
Coins & medals
8%
Jewellery &
silverware
26%
Photography
13% Industrial applications
53%
ETP demand
iShares ZKB physical silver
ETF Securities Silver price (US$/oz)
500 30
450
25
400
350
Mln ounces
20
300
250 15
200
10
150
100
5
50
0 0
Ju -06
Au -06
O -06
ec 6
Fe -06
Ap -07
Ju -07
Au -07
O -07
ec 7
Fe -07
Ap -08
Ju -08
Au -08
O -08
ec 8
Fe -08
Ap -09
Ju -09
Au -09
O -09
ec 9
Fe -09
Ap -10
Ju -10
Au -10
O -10
ec 0
0
D t-0
D t-0
D t-0
D t-0
D t-1
-1
r
r
n
g
n
g
n
g
n
g
n
g
Ap
Source: Credit Suisse, GFMS, Silver Institute
162
THE METALS MARKETS
Figure 6.23 Precious metals: platinum

Platinum mine supply
9,000
Others Zimbabwe North America Russia South Africa
8,000
7,000
'000 ounces
6,000
5,000
4,000
3,000
2,000
1,000
0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Palladium mine supply

9,000 Others Zimbabwe North America Russia South Africa
8,000
7,000
6,000
'000 ounces
5,000
4,000
3,000
2,000
1,000
0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Platinum demand by sector

Other Petroleum Medical & biomedical
10,000 Glass Electrical Chemical
9,000 Jewellery Autocatalyst Investment
8,000
7,000
'000 ounces
6,000
5,000
4,000
3,000
2,000
1,000
0
-1,000
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: Credit Suisse, Johnson Matthey
163
Figure 6.23 Precious metals: palladium

Palladium demand by sector
Other Chemical Jewellery Dental
Electrical Autocatalyst Investment
10,000
9,000
8,000
7,000
6,000
'000 ounces
5,000
4,000
3,000
2,000
1,000
0
-1,000
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Platinum ETF demand

2,250 $2,250
Plat. Ldn Bskt Ldn Plat. ZKB Pt other
2,000 Plat. US Plat. Swiss ABSA Plat, spot

$2,000
1,750
$1,750
1,500
Thousands oz
1,250 $1,500
1,000 $1,250
750
$1,000
500
$750
250
0 $500
Oct-09
Oct-10
Oct-11
Oct-12
Apr-10
Apr-11
Apr-12
Apr-13
Jul-10
Jul-11
Jul-12
Jan-10
Jan-11
Jan-12
Jan-13
Palladium ETF demand

Pall. Ldn Bskt Ldn Pall. ZKB Pd Other
2,500 Pall. US Pall. Swiss Pall, spot $900
$800
2,000
$700
$600
Thousands oz
1,500
$500
1,000
$400
$300
500
$200
0 $100
Oct-09
Oct-10
Oct-11
Oct-12
Apr-09
Apr-10
Apr-11
Apr-12
Apr-13
Jul-09
Jul-10
Jul-11
Jul-12
Jan-09
Jan-10
Jan-11
Jan-12
Jan-13
164
7
Grains and Oilseeds
David Stack
Agrimax
Grains and oilseeds were the first commodities, the staple of our diet
and the basic building blocks for meat and fish (through aquaculture).
In developed economies, food represents some 10% of GDP, higher in
developing economies. Around 20% of people around the world
receive government-subsidised food. This chapter will examine these
crops for each of the major producers and consumers around the
world, analysing how the meat and fish protein markets impact
grains, and the worlds ability to rotate and adjust crop plantings in
the face of a changing demand profile. Likely trends are also noted.
Once the domain of the big grain companies, these commodities
have been a major asset class for investors since the early 2000s, and
this chapter will take a bottom-up approach to analysing the most
relevant information for the various investment themes and their crit-
ical drivers for the years ahead. The traditional power players in the
agricultural markets, both originators and exporters, are the US, the
EU, Brazil and Argentina, and this dominance has been dramatically
affected by the increasing importance in price formation of non-
traditional spheres of influence. Investment themes here have been
greatly influenced by a number of factors, such as the emergence of
China as the worlds largest grain economy while US grain
consumption as ethanol has made it a significantly less important
player for global grains biofuels in general, urbanisation and its
attendant social changes, dramatic changes in food consumption and
rapidly changing agricultural and environmental policy around the
globe.
Surprises that have led to a tightness in these markets include the
165
disappointment of genetic modification (GM) to deliver on its

promise of dramatically improving yields, the slow growth of supply
versus the expected marginal supply curve and the high rate of expan-
sion of Chinas soya demand. There have been the usual droughts,
food scares and a heightened sensitivity to food price inflation, as well
as the global economic crisis and Arab Spring that affected all
investors and markets. However, a number of important players have
entered the agricultural markets. Land investment has become a
mainstream activity, with some spectacular successes and many fail-
ures. The markets evolved from talking about speculation and land
grabs to dividing the new investors into multiple investment styles (as
many as 10, see Appendix 7.1). Some have embraced the traditional,
fundamental style of the agricultural markets, while others intro-
duced new methodologies. Finally, the Dalian Commodity Exchange
became the second-largest futures market in the world, forever
changing the role and dominance of the Chicago Board of Trade
(CBOT). Uniquely, we will examine non-US grain and oilseed
economies; the US is already data-rich and over-analysed, at a time
when its importance in the global grains markets is declining. We look
at the evolution of the Chinese oilseed industry to a staggering 125
million metric tonnes (MMT), far bigger than the US.
This chapter will present an in-depth look at key developments
around the world since the early 1990s, in particular:
the soybean rally of 2003, a surprise for everyone;

the wheat rally of 2007, from sizzling problems to market explo-
sion; and
the maize rally of the 2000s, and the US corn supply/demand net
of ethanol (EtOH).
Finally, we will draw from past market developments to define the

main issues and opportunities for the forward-looking investor.
FEEDGRAINS, FOODGRAINS AND VEGETABLE PROTEINS:

THREE MARKETS, THEIR INDIVIDUAL ECONOMIES AND
INTERDEPENDENCE
The dynamic of these markets lies within the fundamentals, and this
remains the key to understanding them. By first examining the
trends for each of these three markets, and their major producers and
consumers in terms of both the switchable and non-switchable
166
GRAINS AND OILSEEDS
demand and economic drivers, we can then progress to the role of

rotation in their convergence.
The world harvests 2,525 MMT of major grains (corn, sorghum,
barley, wheat, rice, soybeans, rapeseed and sunseed See Table 7.1)
on 800 million hectares (MHas) of land. Across the major producing
countries, the land is devoted as follows: 79% grains (27% maize, 6%
sorghum, 8% barley, 35% wheat and 25% rice) and 21% softseeds
(64% soybeans, 21% rapeseed and 15% sunseed). The average yield is
3.20 MT, of which grains average 3.40 MT and softseeds average 2.20
MT. By crop, the global averages are maize 5.00, sorghum 1.50,
barley 2.63, wheat 3.00, rice 2.88, soybeans 2.50, rapeseed 1.75 and
sunseed 1.50. The gross production tonnages provide the base
volume for each local grain economy, which subsequently
consumes, exports or stores any excess to those two basic needs. We
need to understand the local economy drivers and also the export
availabilities. These exportable volumes, and the extent to which
they are needed in other parts of the world, drive the price as we see
it on the futures markets and through the various cash or physical
prices the commercial world has access to.
AYP is the common industry abbreviation for area in terms of MHa,
yield in metric tonnes per hectare (Mt/HA) and production (the
product of A and Y). In this section, we will discuss the current AYP
for each major grain, where appropriate the whole grain economy for
the major grains, the evolution of the major producer economies since
the early 1990s and their changing role in price formation, as well as
some thoughts on how this may evolve. The following is a summary
of the total of 2,525 MMT of grain production:
850 MMT comes from the nine major maize producers (see Table
(member states of
ndependent States
c of South Africa
rs (see Table 7.3c)
s (see Table 7.3b)
or producers (see
, EU-27, Morocco,
and
167
460 MMT of rice comes from four major producers (see Table 7.7)
Brazil, Thailand, China and India.
Of the soft oilseed total of 350 MMT:
250 MMTof soybeans are from four major producers (see Table
7.9a) US, Argentina, Brazil and Paraguay;
60 MMT of rapeseed are from four major producers (see Table
7.9b) Canada, EU-27, China and India; and
40 MMT of sunseed are from three major producers (see Table
7.9c) Canada, EU-27 and CIS/FSU.
Hard oils (palm oil production) are dominated by Malaysia and

Indonesia at 38MMT (see Table 7.2).
Each major producer has a substantial grain economy for each
grain, and many interact since feedgrains are often combined with
oilseed meals for example, to make complete animal diets. Each of
these economies is different and evolving. There are few clean lines,
with many feedgrains also being foodgrains, and feedgrains being a
major feedstock for biofuels (primarily ethanol, but also sugar cane),
as is vegetable oil (primarily biodiesel).
In Table 7.1 and subsequent tables we compare the last three-year
average of 2010, 2011 and 2012 (201012) to the previous three-year
averages of five years ago (200507), 10 years ago (200002), 15 years
ago (199597) and 20 years ago (199092), to avoid blips in individual
years. We see from this summary that, although the grains area
appears remarkably stable over time (3% 20-year growth), the
oilseeds area has expanded by more than 75%. For combined grains
and oilseeds, the 20-year yield growth has been 24% on an overall
hectare expansion of 13%, leading to a production increase of 40%.
For total arable land, the last five-years production growth came
evenly from area (5%) and yield (6%).
Additionally, we must remember that in this period the US took
around 160,000 km2 or 18.1 MHa (equal to 40 million acres, MAc) out
of production through its Conservation Reserve Program (CRP).
Also, in this period the EU ran its Cereal Set-aside programme. Set-
aside became compulsory in 1992, primarily as a means of reducing
the grain mountain as part of the Common Agricultural Policy. It
was originally set at 15%, before being reduced to 10% in 1996 and
then abandoned in September 2007.
168
Table 7.1 Grain and oilseed 20-year AYP progression comparing three-year averages of 201012 (absolute values) and percentage
growth from 200507 (5-year growth), 200002 (10-year growth), 199597 (15-year growth) and 199092 (20-year growth); MHa,
Mt/Ha and MMT
Crop Last three-year average (201012) 200507 200002 199597 199092
Area A Yield Y Prdn P AYP % growth AYP % growth AYP % growth AYP % growth
Grains 635 3.400 2170* 2% 8% 10% 6% 16% 23% 2% 23% 25% 3% 28% 31%**
Oilseeds 164 2.200 355 15% 3% 18% 34% 8% 45% 52% 20% 82% 78% 27% 125%
Total arable 799 3.200 2525 5% 6% 11% 11% 13% 26% 9% 20% 31% 13% 24% 40%
Source: Adapted from Informa; * 635/3.40/2,170 means grains area is 635 MHa, world average yield is 3.40 Mt/HA and world production is 2,150
MMT; ** 3%/28%/31% means grains area has grown 3% in 20 years, yield has grown 28% and production by 31%
GRAINS AND OILSEEDS

169
Table 7.2 Major exporters and major importers, by grain or vegetable protein
Corn/Maize Wheat Soybeans/Meal/Oil Rapeseed Palm Oil

100 MMT 130 MMT 93/55/8 MMT 11MMT 38 MMT 70%
of the 54 MMT
of global veg
oil
Top 10 Exporters
1 US US US/Arg/Arg Canada Indonesia (19.0)
2 Argentina Australia Brz/Brz/Brz Australia Malaysia (18.7)
3 Ukraine Canada Arg/US/US Ukraine
4 Brazil EU-27 Paraguay/India
5 India Russia Canada/China/
6 Russia Argentina
7 RSA India
8 Paraguay Ukraine
9 Canada Kazakhstan
10 EU-27 Turkey
Corn/Maize Wheat Soybeans/Meal/Oil Rapeseed Palm Oil
Top 10 Importers
1 US EtOH Egypt China/EU/China EU-27 India
2 Japan Brazil EU27/Indonesia/
India Japan China
3 EU-27 Indonesia Mexico/Vietnam/
Iran China EU27
4 Mexico Japan Taiwan/Thailand/
Bangladesh Mexico Pakistan
5 South Korea Algeria Japan/Japan/
Venezuela US Singapore
6 Egypt South Korea Thailand/Philipp/
Peru Canada Egypt
7 Iran Mexico Indonesia/Iran/
Algeria US
8 Taiwan Iraq Egypt/South Korea/
Egypt Bangladesh
9 Colombia Morocco US/Mexico/South
Korea CIS/FSU
10 Algeria Nigeria/Philipp. South Korea/Canada
& Colombia/
Morocco & RSA Iran, Vietnam
& Japan
FEEDGRAINS
There are three major feedgrains in the world corn, sorghum and
barley the production of which amounts to 1,045 MMT. However,
we must add significant volumes of feed wheat consumed in China
(1012 MMT), the EU (4957 MMT in 200712), Russia (13 MMT) and
170
GRAINS AND OILSEEDS
India (34 MMT), a total of 80 MMT. It is important to remember that

significant quantities of feed wheat produced globally is fed to
animals the broad consensus says this is about 17% of global wheat
production, or 115 MMT.
The world harvests 855 MMT of maize annually, whose major
economies are the US (300), Argentina (25), Brazil (65), Mexico (20),
France (15), EU-27 (60), CIS/FSU (30), RSA (10), Thailand (5) and
China (195). Grain sorghum production of 60 MMT, widely distrib-
uted around the world, has major production in the US (7),
Argentina (5) and Australia (2). For both of these crops, there is also
significant non-grain production as feed, used as silage best
described as a whole, above-ground crop, whose stems, leaves and
grain ear are pickled in vinegar (eg, formic acid) to preserve it, before
it is stored and fed to livestock over the following winter. Barley
totals 130 MMT, of which Canada (10), EU-27 (55) and CIS/FSU (25)
are the major economies. Note that, at 130 MMT, barley production is
greater than Chinas wheat production (second only to EU wheat
production) and twice that of Brazils corn production (the worlds
third-largest corn producer). Both barley and sorghum are in decline
in terms of devoted area and the world barley and feed-wheat
markets are the same size.
The major feedgrains are starch or carbohydrate producing and
consumed by animals, hence the feed designation. They also have
considerable industrial use. We can divide the animal kingdom into
two stomach types: monogastric and ruminant. We humans are
monogastric, having one simple stomach, as are pigs and
chickens, while cattle are ruminants, having a rumen. The rumen
can be thought of as a vat, capable of stewing and digesting highly
fibrous food, such as grass and leaves, which contain carbohydrates
bound by lignin, a complex fibre. Feeds such as potatoes require
boiling to break down their complex carbohydrate structure to
make them easily digestible for monogastrics. Grains are simply
processed by grinding to break down the husk or outer covering,
rendering them easily digestible to a ruminant, while full milling
and husk removal makes them also easily digestible by mono-
gastrics. As the reader will be well aware, there is an ongoing
conflict between ease of digestibility and the many essential nutri-
ents found in the husk wholegrain bread being the classic
compromise for humans.
171
Table 7.3a Maize 5-, 10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 199092).
After each country name the % of world area devoted to this crop in 2012 and 1992 are given
Last 3 year average
20102012 20052007 20002002 19951997 19901992 % Of world
Area A Yield Y Prdn P AYP % growth AYP % growth AYP % growth AYP % growth yield
Maize / Corn
US 35% 21% 34.1 8.834 301 9% 6% 2% 20% 5% 26% 20% 15% 38% 22% 18% 43% 178%
Argentina 3% 2% 3.7 6.625 24 28% 1% 26% 44% 12% 62% 19% 35% 59% 63% 59% 158% 133%
Brazil 8% 10% 14.4 4.526 66 4% 25% 30% 15% 41% 62% 11% 79% 100% 9% 114% 133% 90%
Mexico 2% 5% 6.5 3.087 20 8% 1% 8% 11% 18% 5% 16% 34% 13% 7% 38% 27% 63%
France 2% 1% 1.6 9.347 15 6% 4% 10% 13% 5% 8% 7% 12% 3% 8% 32% 21% 188%
EU 27 7% 3% 8.6 6.870 59 1% 11% 9% 6% 14% 7% 24% 0% 35% 97% 3% 89% 138%
CIS/FSU 3% 2% 5.9 4.734 28 58% 31% 110% 120% 66% 269% 130% 64% 271% 108% 51% 216% 95%
South Africa 1% 3% 3.1 3.953 12 13% 20% 34% 11% 48% 33% 18% 61% 32% 8% 82% 68% 80%
Thailand 1% 1% 1.0 4.248 4 1% 13% 12% 13% 11% 3% 11% 28% 15% 22% 55% 20% 85%
China 23% 16% 33.3 5.779 193 19% 10% 30% 39% 22% 69% 41% 19% 68% 56% 27% 99% 115%
World total 100% 100% 168.6 5.061 853 11% 5% 16% 23% 16% 43% 22% 24% 52% 29% 32% 69% 100%
Table 7.3b Barley 5-, 10-, 15- and 20-Year AYP Progression (comparing 2010-12 with 2005-07, 2000-02, 1995-97 and 1990-92).
Last 3 year average
20102012 20052007 20002002 19951997 19901992 % Of world
Barley
Canada 5% 5% 2.5 3.125 8 30% 5% 27% 37% 20% 25% 46% 4% 44% 40% 9% 35% 119%
EU27 25% 20% 12.5 4.250 53 11% 6% 5% 12% 0% 12% 17% 3% 15% 15% 4% 12% 162%
CIS/FSU 27% 37% 13.5 2.000 27 18% 7% 12% 17% 2% 16% 42% 39% 18% 50% 10% 45% 76%
World total 100% 100% 49.5 2.625 129 12% 8% 5% 10% 5% 6% 26% 16% 14% 33% 13% 25% 100%
Table 7.3c Sorghum 5-, 10-, 15- and 20-Year AYP Progression (comparing 2010-12 with 2005-07, 2000-02, 1995-97 and 1990-92).
Last 3 year average
20102012 20052007 20002002 19951997 19901992 % Of world
Sorghum
172
US 5% 10% 1.9 3.713 7 21% 10% 30% 41% 4% 39% 53% 8% 57% 56% 9% 60% 243%
Argentina 3% 2% 1.1 4.367 5 87% 7% 74% 93% 12% 70% 53% 11% 67% 52% 18% 78% 286%
Australia 2% 1% 0.7 3.181 2 12% 11% 7% 11% 36% 20% 20% 33% 59% 46% 64% 133% 208%
World total 100% 100% 38.5 1.527 59 8% 4% 4% 5% 3% 3% 9% 5% 4% 3% 5% 2% 100%
GRAINS AND OILSEEDS
For many animals, grains are a simple supplement for their diet
(eg, beef cattle that consume mainly forage), while dairy cows, pigs
and poultry require a considerable amount of protein to be added to
their diet since they perform optimally with a 2025% protein feed,
almost twice that of any cereal grain. This implies a 75:25 grain:meal
combination. Therefore, major feedgrain consumers must also
produce or import their protein needs, making the EU a major
importer of softseed proteins. Although it may seem very straight-
forward, the Pearson Square formulation works surprisingly well for
estimating diets for forecasting animal or aquaculture needs, and is
easily found with a web-search.
Biosystems, of which one is the stomach, are complex and a series
of associative effects can be observed. We do not digest equally meat
and potatoes that are eaten separately, as compared to eating combi-
nations in various proportions. The cooking method and previous
meal also influence digestion. We do not similarly digest meat and
rice in the same way as meat and potatoes. This leads to feed conver-
sion efficiency (FCE), a metric which is the first step towards
metabolisability, the rate at which we actually use the nutrients we
have ingested. FCE is normally expressed as kilograms (kg) of dry
matter output per Kg of DM feed.
In principle, as we allocate raw materials, feedgrains should only
go to those processes that efficiently transform them into human
food. For example, this means that we would not feed grains to cattle
other than what is required to optimise their ability to digest cellu-
losic feeds. If this were the case, and we were simply economic
actors, we would have more than enough to feed the world
however, this would have the effect of large parts of the common diet
disappearing around the world. We prefer to eat as we please,
dependent on prevailing price and income.
Associative effects include the reality of optimised nutrition,
combining carbs, proteins and fats to get the optimal feed conver-
sion. Diets have been balanced at the commercial level based on the
least-cost formulation since the 1960s, and it remains a simple linear
programming exercise. Animal nutrition has advanced much faster
than human nutrition, not least because we can isolate genetics and
enforce diets for animals, before butchering them to measure the
output much more easily than with humans. Optimising nutrition is
scientifically easy but socially complex, and we can imagine very
173
different optimal strategies for an Olympic athlete and a couch

potato, or a new born infant and an octogenarian. Calories are to
modern nutrition what gasoline was to the Model T Ford: raw
energy. We have come to think in terms of metabolisability (useful-
ness) of carbs, proteins and fat, and also the various micronutrients
and salt balances that influence our bodies and lives.
Wheat and barley belong to a category of grains (which includes
oats) that the US Department of Agriculture (USDA) refers to as the
small grains, and which are reported separately. A report issued
annually in December and various updates provide detail on a state-
by-state basis of the AYP of these three crops. The EU treats wheat in
the same way that the US does maize, since it is the base of animal
feed, and consequently describe everything else as coarse grains.
The reader must be careful to compare coarse grains in different
grain economies they mean different things.
There are over a dozen major feedgrain economies in the world,
all growing grains and other food and feed in rotations customised
for the location, growing degree days (GDDs) and current
economics. A major grain economy is defined by the author as one
producing more than several million tonnes in excess of its local
requirements through rotation. An example is Ukraine, which
produces roughly 0.5 tonnes of wheat per capita. Its enormous
simultaneous local production of potatoes leaves it with a huge
exportable surplus of basic carbohydrate. The US produces one
tonne of corn for every inhabitant. Once you get in the grain or
oilseed producing business as a farmer, the quality of your output
and its ultimate designation as food or feed will depend on the
variety you chose to plant, how you cared for it, mother nature
(weather), evolving global demand, the market where you choose to
sell it and the degree to which it is carefully handled, processed,
marketed and blended.
In the early 1990s, the US dominated the maize market globally as
a producer and exporter. Its exports were residual to its own animal
feed and food, seed and industrial (FSI) needs, and it carried large
stocks. In 1990/91, the US had almost 35 MMT in stock, producing
200 MMT and exporting 55 MMT. The US Maize crop year (CY)
begins in September and ends before or at the start of harvest in
August of the subsequent year. Optimal planting is between April
1st and May 30th while harvest runs from August 20th to November
174
Table 7.4 True US corn (maize) annual S&D (MMT), using 1990/91 crop year alcohol as base: a 20-year perspective (September
August crop)
Crop year 1990/91 1995/96 2000/01 2005/06 2010/11 2011/12 2012/13
US (September/August)
Harvested Area (MHa) 27.1 26.4 29.3 30.4 33.0 34.0 35.5
Yield (MT/Ha) 7.44 7.12 8.59 9.29 9.59 9.24 7.67
Carryin 34.2 39.6 43.6 53.7 43.4 28.6 25.1
Carryin less 20 days 'fuel as corn' 34.2 38.8 42.8 51.9 37.0 21.7 18.1
Production 201.5 188.0 251.9 282.3 316.2 313.9 272.4
Production less EtOH 201.5 186.8 244.8 250.5 197.6 195.5 164.5
Imports 0.1 0.4 0.2 0.2 0.7 0.7 3.2
Total supply 235.8 228.0 295.7 336.2 360.2 343.3 300.7
Adj total supply (ex fuel) 235.8 226.8 288.6 304.4 241.6 224.9 192.8
Use
Feed & Residual 117.1 119.4 147.9 155.3 121.7 115.5 109.2
% Adj Tot Supply 50% 53% 51% 51% 50% 51% 57%
Food/Seed/Ind 36.2 41.4 50.2 76.7 163.3 163.5 153.0
Ethanol FSI 8.9 10.1 16.0 40.7 127.5 127.3 116.8
Fuel FSI 0.0 1.2 7.1 31.8 118.6 118.4 107.9
Non Fuel FSI 36.2 40.2 43.1 44.9 44.7 45.1 45.1
Total FSI as % total supply 15% 18% 17% 23% 45% 48% 51%
Non fuel FSI as % total supply 15% 18% 15% 15% 18% 20% 23%
Adj domestic use (ex fuel) 153.3 159.6 191.0 200.2 166.4 160.6 154.4
Exports 43.9 56.4 49.3 54.2 46.6 39.2 25.4
Exports as % adj total supply 19% 25% 17% 18% 19% 17% 13%
Exports as % adj domestic use 29% 35% 26% 27% 28% 24% 16%
GRAINS AND OILSEEDS

Adj total use 197.1 215.9 240.3 254.4 213.0 199.8 179.8
Carryout 38.6 10.8 48.2 50.0 28.6 25.1 13.0
Adj Carryout (ex 20 days 'fuel as corn') 38.2 10.3 47.4 47.7 21.7 18.1 6.6
Adj C/O as % adj domestic use 25% 6% 25% 24% 13% 11% 4%
20 days of fuel as corn 0.5 0.6 0.9 2.2 7.0 7.0 6.4
175
Source: Agrimax
Note: In each step the traditional USDA format is improved by deducting maize produced for fuel EtOH and the appropriate stocks deducted.
30th. Both run progressively northwards. The current crop was

planted two weeks late. The US plants 353.5 MHa, and has a three-
year average yield of 8.875 MT/Ha, which includes the disastrous
harvest after the drought of 2012. Yields are normally expected to be
over 9.0.
Since the two major drivers of domestic maize consumption for all
major producers are Feed and FSI, Table 7.4 takes the unusual
approach of stripping out the corn-for-ethanol maize demand to
allow accurate comparison with other countries. We can then
proceed to examine the US and other maize-economies sequentially
and draw some conclusions for the future. Table 7.4 follows usual
USDA protocol so Area and Yield are directly comparable with
USDA. Thereafter, line by line, it strips out the ethanol demand
which is not a grain demand and allows us to get to non-fuel FSI by
freezing fuel ethanol demand at 1990/91 CY-levels and shows in the
Adjusted domestic use row that demand is in fact almost flat in the
US from 1990 CY to 2013.
The feed economy

One way to quantify grain demand is to employ feed-use data and
grain-consuming animal units (GCAUs), factors that allow compar-
isons of grain demand among different types of livestock. One
GCAU is 2.15 tons (short tons have 2,000 pounds, while metric have
2,204.6). The USDA has developed a different factor for each type of
livestock based on the average amount that one such animal
consumes in a year. For example, a dairy cow has a GCAU factor of
1.0474, while a broiler has a factor of 0.002. Using these factors, we
can see that one dairy cow will use the same amount of grain (1.0474
2.15 tons = 2.25 tons) in a year, as approximately 523 broilers (one
broiler will consume 0.002 2.15 tons = 0.0043 tons, and 2.25 divided
by 0.0043 equals 523 broilers). The major GCAU factors are feeder
cattle: 0.0547, broilers: 0.002, layers: 0.0217, turkeys: 0.0155, dairy
(cow + calf): 1.0474 and hogs: 0.2285. Informa Economics, Inc. offers
the best analysis of GCAUs and also protein-consuming animal
units (PCAUs), allowing us to view the relative intensity by animal
type of each major feed component side by side.
Globally, we are eating an increasing amount of white meat,
resulting in greater numbers of monogastrics and increased feed
conversion efficiency (FCE). Two important issues arise here: there
176
GRAINS AND OILSEEDS
has been a dramatic increase in industrial (large-scale) animal

farming since the early 1990s, as well as further urbanisation, which
are additive in effect. The effort to supply large quantities of meat
requires significant supply chains and consumer packaged goods
(CPGs) provided by companies such as Nestl and Danone, meat
companies and retail supermarkets.
Animal feed demand is not hardwired in the same way as FSI for
two reasons. The feed compounder can choose many feeds to make
the ration, and the consumer has a lot more discretion, cut by cut, as
to what meat they choose to eat. It is beyond the scope of this chapter
to discuss global meat demand, but we do size the larger food
protein economies US, China, EU, Russia, Brazil and India (Table
7.5) and look at the broad consumption figures. In addition, we
note that, in much of the world, consumers will switch between
different food proteins as their relative price changes. Price changes
for proteins are frequently more volatile than for grains or oilseeds.
Pork accounts for 60% of Chinas meat protein consumption. In
general, poultry is substituted as a meat protein when pork prices
reach high levels. Conversely, when pork prices are affordable,
Chinas consumers prefer to purchase pork products.
Before looking at animal feed, we should briefly review animal
protein consumption. For some inexplicable reason, it is unusual to
find this critical information in most discussions on grains and
oilseeds. Table 7.5 shows that China leads on production and
consumption, consuming twice as much meat as the US, while in 1990
they consumed roughly the same. In terms of quick numbers, this
means the average Chinese person eats half as much meat protein as
the average American, 40% more pork per capita, one quarter as much
chicken and one ninth as much beef. From a tiny chicken industry in
the early 1990s, China has come to consume more chicken than the US,
at some 14 MMT. Not only does China consume 33% of the worlds
meat, but also 33% of the worlds fish and aquaculture, and in 2010 it
became the largest animal feedgrain user, including an estimated 12
MMT of wheat. In terms of total animal protein, China is twice as big a
consumer than both the EU 27 and US.
Outside of China, Brazil has become a major meat exporter. In
addition, despite being widely thought of as a vegetarian country,
India consumes almost 20 MMT of meat per annum. It is estimated to
consume 5.56.0 kgs per capita of chicken with a retail value of
177
US$9.0 billion, and it is widely touted to become more populous than

Chinas 1.35 billion people. Therefore, the forecasted 20% growth in
chicken consumption in India will have an impact on the grain
markets. Russia is seriously underserved in the animal protein cate-
gory. In the not too distant future, the author would expect the
Chinese and Brazilian poultry economies to surge past the US and
EU markets. As a final caution against analysing enormous popula-
tions, remember the ag majors dig a lot deeper into this kind of data,
and categorising 100 million people never mind 1.35 billion as
behaving in a homogenous way is intuitively risky.
The (FSI) industrial corn economy

To compare FSI of the major producers we must use the adjusted
true corn supply and demand (S&D) for the US. It has a 45 MMT FSI
demand, roughly half the size of its feed consumption, growing 25%
in 20 years and from 15% to 26% of the adjusted production which
makes it globally comparable. Argentina consumes 2.2 MMT (up
90% in 20 years, yet declining from 14 to 8% of the crop); Brazil
consumes 7.0 MMT (up 100% and down from 14% to 11% of the
crop); Ukraine uses 1.5 MMT (up 33%, from 21 to 7%); Russia 1 MMT
(down 40%, from 44 to 10%); the EU 15.5 MMT (up 45%, from 30 to
28%); RSA 4.5 MMT (up 15%, from 46 to 35%); China 64 MMT (up
100%, from 27 to 31%) and India 8.3 MMT (up 23%, from 78 to 42%).
Non US major producers total 110 MMT in FSI, an important 2.4
times the US, growing 84% over 20 years and declining only slightly
from 28 to 25% of local production. The industrial corn economy is
aimed at high value-added processing, and a typical analysis is
heavily clouded by the conventional reporting process of FSI, which
includes fuel ethanol. FSI has no meaningful seasonality while feed
demand does.
There are two main types of corn processing: dry milling (EtOH)
and wet milling (sweeteners). The products of each type are utilised
in different ways. Over 80% of US ethanol is produced from corn by
the dry milling process. The ethanol is dehydrated to about 200
proof using a molecular sieve system, and a denaturant such as gaso-
line may be added to render the product undrinkable. With this last
addition, the process is complete and the product is ready to ship to
gasoline retailers or terminals. The remaining stillage then under-
goes a different process to produce a highly nutritious livestock feed
178
GRAINS AND OILSEEDS
(DDGS). The carbon dioxide released from the process is also utilised
to carbonate beverages and in the manufacturing of dry ice. Ethanol
yield is constantly rising and water use efficiency improving. The
initial assumption that biofuels were good for the environment
because they had a smaller carbon footprint is debatable regarding
the contention that the production of grain alcohol, and therefore
E15, may actually have a greater environmental impact than fossil
fuels.
US non-fuel FSI averaged 15% of production across the 20-year
period, amounting to some 42 MMT. In theory, this all comes from
wet milling, a process which takes the corn grain and steeps it in a
dilute combination of sulphuric acid and water for 2448 hours in
order to separate the grain into many components. The slurry mix
then goes through a series of grinders to separate out the corn germ.
This process is the backbone of industrial processing for the produc-
tion of fructose, glucose, dextrose, starch, potable alcohol and
industrial alcohols. These figures are typical of an industrial maize
economy found all over the world with the exception of high-
fructose corn syrup (HFCS) and fuel ethanol, which are US-specific.
In 20 years, US production of HFCS increased by 33%, glucose and
dextrose by 54%, starch by 14%, potable alcohol was unchanged and
cereal consumption increased by 64%, largely driven by the USDA
food pyramid. The growth is predictable since the plants are
announced and take time to build. This industrial demand is largely
non-switchable. For example, it was affected by a 2006 agreement
(which became effective in 2008) to allow sweeteners to flow from
the US to Mexico without tariffs.
HFCS is produced by wet milling corn to produce corn starch,
then processing that starch to yield corn syrup, which is almost
entirely glucose, and then adding enzymes that change some of the
glucose into fructose. The resulting syrup (after enzyme conversion)
contains naturally 42% fructose, and is consequently called HFCS 42.
Some of the 42% fructose is then purified to 90% fructose (HFCS 90).
To make HFCS 55, the HFCS 90 is mixed with HFCS 42, and this
increased fructose percentage gives it the same sweetness taste as
sugar (which is why it is called high fructose corn syrup).
A system of sugar tariffs and sugar quotas imposed in 1977 in the
US significantly increased the cost of imported sugar, and US manu-
facturers therefore sought cheaper sources. HFCS, as it is derived
179
Table 7.5 FAO estimates of world 2010 animal protein consumption by type major economies (MMT)
Beef and Pork Poultry Meat Commercial catch Total %Meat %of world
veal of world & meat protein in diet fish
aquaculture
World 57.3 103.2 76.0 236.5 142.0 100% 615.0 38% 100%
US 12.0 10.2 16.5 38.7 4.9 16% 82.3 47% 3%
EU-27 8.1 23.0 9.0 40.1 6.4 17% 86.6 46% 5%
Brazil 9.1 3.2 12.3 24.6 0.5 10% 49.7 49% 0%

Russia 1.4 1.9 2.3 5.7 3.5 2% 14.8 38% 2%
India 2.8 2.6 5.5 7.5 2% 18.4 30% 5%
China 5.6 51.1 12.5 69.2 47.5 29% 185.8 37% 33%
Top ten consumers by rank

1 US China US China
2 Brazil EU 27 China India
3 EU-27 US Brazil Peru
4 China Brazil EU 27 Indonesia

5 India Russia Mexico US
6 Argentina Vietnam India Japan
7 Australia Canada Russia Chile
8 Mexico Japan Argentina Vietnam
9 Pakistan Philippines Iran Thailand
10 Russia Mexico Thailand Russia
180
GRAINS AND OILSEEDS
from corn, is more economical because the domestic US prices of

sugar was twice the global price while the price of corn was kept low
through government subsidies to growers. HFCS became an attrac-
tive substitute, and was preferred over cane sugar by the vast
majority of US food and beverage manufacturers. Soft drink makers
such as Coca-Cola and Pepsi use sugar in other countries, but
switched to HFCS in the US during the mid-1980s. In 2010, the Corn
Refiners Association applied to allow HFCS to be renamed corn
sugar, but this was rejected by the US Food and Drug
Administration in 2012.
Barley (Hordeum vulgare L.) is a member of the grass family and
therefore closely related to wheat, and is a major cereal grain.
Important uses for barley are as animal feed, as a source of
fermentable material for beer and certain distilled beverages, and as
a component of various healthfoods. It is used in soups and stews,
and in barley bread. Malting barleys are normally separate and
distinct varieties from feed barley. In a ranking of cereal crops in the
world, barley is fourth, both in terms of quantity produced and area
of cultivation. For our purposes we include it in feedgrains although,
as with most of these crops, the lines are blurred.
Canada, the EU and CIS/FSU are the major barley producers (see
Table 7.3b), accounting for 70% of global production, and their yields
are quite different at 3.125, 4.25 and 2.00 MT/Ha, respectively, giving
very different competing crop economics. As one can imagine, the
decline in area has been greatest in the low yielding producers, and
in 20 years Russia fell from almost 50 MMT to almost 25, and global
production decreased from 170 to 130 MMT, down some 33%, of
which the big three declined by 40, 15 and 50%. Barley has been
closely associated with small farms and on-farm feeding, which
means the decline will continue.
Sorghum is in a genus of numerous species of grasses and a rela-
tive of other C4 plants like maize and sugarcane. With lower yields
than maize the US (the world's largest producer, see Table 7.3c) has
more than halved its sorghum crop to 10% of global production, the
remainder being scattered around the world where it may be locally
important as food or feed. Many species are cultivated in warmer
tropical climates worldwide. It is also biologically in the same tribe
and subfamily as sugarcane and might have been grown widely in
Brazil where local tastes prefer rice as food carbohydrate. Globally its
181
Table 7.6 Wheat five-, 10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 199092).
Wheat Last three-year average (201012) 200507 200002 199597 199092 % of world
US 9% 11% 19.0 3.000 59 4% 13% 9% 3% 16% 12% 24% 22% 7% 25% 21% 10% 100%
Canada 4% 6% 9.0 2.875 25 5% 12% 7% 13% 39% 20% 24% 27% 4% 37% 29% 19% 96%
Argentina 2% 2% 4.0 3.625 15 29% 29% 8% 37% 59% 0% 29% 60% 14% 16% 70% 46% 121%
Brazil 1% 1% 2.0 2.500 5 6% 41% 48% 21% 72% 107% 44% 60% 128% 15% 100% 78% 83%
EU 27 12% 8% 25.5 5.250 135 3% 4% 7% 2% 7% 4% 10% 12% 24% 35% 2% 32% 175%
Morocco 1% 1% 3.0 1.625 5 4% 28% 29% 10% 58% 75% 22% 38% 51% 19% 21% 39% 54%
CIS/FSU 22% 21% 49.0 1.875 92 5% 2% 3% 9% 0% 9% 4% 29% 35% 4% 0% 4% 63%
Turkey 4% 4% 8.0 2.125 17 6% 6% 0% 9% 12% 2% 8% 17% 8% 11% 20% 7% 71%
China 11% 14% 24.0 4.875 118 4% 8% 12% 3% 29% 24% 18% 28% 5% 21% 51% 19% 163%
India 13% 10% 29.0 3.000 87 8% 13% 22% 11% 8% 20% 14% 16% 33% 23% 32% 63% 100%
Australia 6% 4% 13.5 2.000 26 11% 45% 60% 14% 24% 40% 31% 3% 35% 60% 19% 90% 67%
World total 100% 100% 220.5 3.000 668 2% 7% 10% 2% 13% 15% 2% 19% 16% 2% 21% 18% 100%
182
GRAINS AND OILSEEDS
stagnant 60 MMT production is neither important in world trade nor

expected to be so.
FOODGRAINS
There are two major foodgrains: wheat and rice. The world harvests
670 MMT of wheat annually (see Table 7.6), of which the major
wheat economies are the US (60), Canada (25), Argentina (15), Brazil
(5), EU-27 (135), Morocco (5), CIS/FSU (90), Turkey (15), China (120),
India (85) and Australia (25). World rice harvests is 460 MMT (see
Table 7.7), of which Brazil (10), Thailand (20), China (140) and India
(100) are the largest economies. There are thousands of wheat vari-
eties being grown in the world, each selected, bred and adapted
based on locality and consumer preference.
Table 7.6 shows that US wheat production is flat (area declines
and yield improves) and expected to decline as maize takes up more
land for ethanol. Canada, Argentina, Australia and Brazil are stag-
nant, while the EU, China and India have grown quite dramatically.
In addition, the FSU declined dramatically as it became more
market-based, but has considerable potential to recover production
through the use of modern farming methods. Yield growth around
the world remains good, in many cases due to suboptimal wheat
areas being taken out of production in China and the FSU.
Throughout the world, there are various ways of categorising
wheat, largely dependent on intended use. We can think of wheat
globally and genetically as having 10% protein content, often
referred to as its fair merchantable quality (FMQ). FMQ changes
with variety, husbandry and weather. While the EU tends to specify
wheat by specific weight (in the US, it is thousand grain weight,
TGW) measured in kilograms per hectolitre (Kg/hl) and variety,
feed wheat is generally assumed to have a 72 kg/hl FMQ (UK Liffe
contract spec) and milling or baking wheat to have a 76 kg/hl FMQ
(French Euronext contract spec).
The most common simple laboratory test for protein quality
(gluten) is the Hagberg falling number (HFN), which measures the
rate of fall of a plunger through a column of water/flour mix, repre-
senting its stickiness or so-called gluten extensibility the ability of
the wheat to form a uniform rising dough. From the most simple
feed/food designation in Europe, each major wheat exporter has its
own preferred designations. A wheat chapter that does not discuss
183
type by geography and consumption would be pointless, so we will

take a deeper look at China and India. Wheat varieties can vary over
100 miles, and there are thousands around the world.
In fact, there are several hundred varieties of wheat grown just in
the US, all of which fall into one of six recognised classes. Wheat
classes are determined not only by the time of year they are planted
and harvested, but also by their hardness, colour and the shape of
their kernels. Each class of wheat has its own family characteristics,
as related to milling and baking or other food/feed use. Wheat
production by type across the states and then subsequently for
spring, winter and durum, and the intensity by county within each
state can be seen at: http://www.thefreshloaf.com/node/4632-
/major-wheat-growing-regions-us-reference-maps.
The largest volume of US wheat is of Chicago Board of Trade
(CBOT) type and specification and is often referred to as simply W.
CBOT-type wheat is both an animal feed and capable of making
biscuit dough, or a simple unleavened dough, and has low protein
content and poor gluten extensibility. Kansas City Board of Trade
(KCBT) wheat (often referred to as KW) is true bread wheat destined
for human consumption but capable of being fed in small quantities
with other grains to animals. Its gluten extensibility is sufficient to
capture a bubble of air and allow the dough to rise to produce a loaf
of bread. Minneapolis Grain Exchange (MGE) wheat (often referred
to as MW) is best thought of as a high-class or technical wheat
capable of making fine pastries such as croissants. The gluten is
extremely flexible and can produce a low-dough large bubble.
W has no protein minimum per se, while KW has 11% and MW has
13.5%. All three futures contracts are based on #2 grade, which is a
minimum TGW, #1 would be higher and sub-economic to deliver as
we can simply blend it down. All of these markets carry a variety of
scales that adjust for delivering #3 grain (lower specific weight), and
KW and MW allow for penalties to be deducted for protein levels
down to 10.5% and 13.0%, respectively. Protein scales refer to the
per 0.5% or 1.0% value for protein quoted in the physical or cash
markets, and represent the value of different grades that are blended
by millers and shippers to make actual grists (the bakers wheat
slate) or shipping contract minimums. Protein levels do not blend
linearly but are close enough for anything we need to discuss here.
Baking is in fact a science, and there is a large body of work available
184
GRAINS AND OILSEEDS
on wheat qualities, baking and the use of gluten-extenders, for

example, a man-made additive intended originally for use in poor
harvests but now used widely to create uniformity.
Although many talk about declining wheat per capita consumption
and the rise of corn and soybean production, this masks a much more
complex picture. To talk of wheat and bread is a serious mistake,
demographically. The grains and oilseeds industries of other coun-
tries are substantially different to the OECD; Mexico thinks about
tortillas more than loaves of white bread, the Indian/Pakistan wheat
economy is very large but flour production and sales direct to the
consumer, rather than bread, remains a substantial industry, while
Asian noodles are a major source of wheat consumption all over the
globe, not just in Asia. Although much has been written on wheat and
foodgrains, most focus on the easily researched OECD producers: US,
Canada (three planting zones, 14 classes and three or four grades for
export of each variety), Australia (five planting zones, six principal
grades targeting 13 end uses, from Indian bread to Udon noodles and
Asian instant noodles), Argentina (seven planting zones, three major
categories, four flour grades) and France (17 planting zones or areas,
four classes and four grades, all variety-specific). We will look at the
two most populous countries, China and India, to provide a cross
reference of their enormous wheat economies rarely found outside an
ag major or the most serious investor. Without understanding these
two rapidly evolving wheat economies there should be little expecta-
tion of understanding price evolution.
Prior to the expansion of the EU to 27 countries, China was the
worlds largest producer and consumer of wheat. Comparative
advantage has led China to discourage low-quality wheat produc-
tion, and it has reduced the amount of land devoted to wheat since
the early 2000s. It imports as a way of balancing quality not quantity,
and the US wheat class designations will not advance your under-
standing of Chinese wheat needs. Its planting zones can be divided
broadly into three: hard wheats around the Greater Khingan
Mountain range, hard wheats along the Yellow, Huai and Hai rivers
and, finally, soft wheats along the lower Yangtze river. They include
nine classes (the first is H or S for hard and soft, then W or R for white
or red and W or S for winter or spring the main ones are HWW,
HWS, SWW, SWS, HRW, HRS, SRW, SRS and other), and five grades
(79+, 77+, 75+, 73+ and 71+ Kg/Hl) by specific weight and a variety
185
of other quality characteristics, including moisture and foreign

matter.
The nine Chinese grades are further divided into two classes of
high-quality strong gluten wheat, two classes of high-quality weak
gluten wheat and three other qualities for specific end-uses. Each
grade class has a specific flour quality, including HFN and a range of
other specific qualities. In addition, each of the classes and flour
grades are identified by planting zone. Some 10% of the wheat is
used for high-quality bread cookies and dumplings, some 50% for
steamed bread, noodles and instant noodles, and the remaining 40%
is used locally for home baking or small bakeries. Urbanisation and
the industrialisation of its food industry is dramatically changing the
patterns of consumption in China. Despite their large total animal
protein consumption, the population effect means China (and
indeed Asia) depends heavily on foodgrains for nutrition.
Noodles represent some 40% of total flour consumption, and are a
major staple in East and Southeast Asian countries. Apart from
wheat flour, they can be made from rice flour, potato flour, buck-
wheat flour, corn flour, bean, yam and soybean flour. While pasta is
made from tetrapolid durum wheat (Triticum durum), noodles are
made from the hexaploid Triticum aestivum, which contains gluten,
which reacts to the pressure during the sheeting process. Eggs are
frequently added to provide a firmer texture.
Given that wheat consumption in the form of Asian noodles
exceeds the total US wheat production, we can understand its signif-
icance in the forecasting of demand for wheat round the world.
Chinese noodles are typically made from hard wheat flours,
Japanese noodles from soft wheat of medium protein. By colour,
they are typically classified as white (containing salt) or yellow
(containing alkaline salt). White salt noodles include Japanese
noodles, Chinese raw and dry noodles. Yellow noodles include
Chinese wet noodles, Hokkien noodles, Cantonese noodles, Chukka-
men, Thai bamee and instant noodles. Over 50 billion meals are
annually served around the world that contain ramen noodles alone.
Asia imports US HRS, DNS, HRW, SRW and SWW, Australian stan-
dard white (SW), premium white (PW) and prime hard wheats (PH),
as well as Canadian Western Red Spring (so called CWRS), Canadian
Western Red Winter, Canadian Prairie Spring White and Canadian
Prairie Spring Red wheats to blend with local wheats to make
186
GRAINS AND OILSEEDS
noodles. China consumes 35% of global instant noodles, just twice as

much as Indonesia, which is twice that of Japan. The US and South
Korea consume one fifth as much as China.
Steamed bread accounts for 60% of flour consumption in
Northern China (where it is a staple) and 2030% in the South (where
it is a dessert). In Asia, it represents 515% of flour consumption
depending on the country, and it is popular in the Philippines, for
example. It is made predominantly from soft-to-medium hard
wheats and, while it is prepared somewhat like a western pan bread,
it is then steamed rather than baked. There are three principal types
with varying protein and gluten qualities Northern, Southern and
Taiwanese. The Northern type is typically made from local wheat,
and has 1011% protein. The Southern type has added sugar and
baking powder. The Taiwanese type has the highest protein, while
all three types contain yeast. The steaming process produces a
higher-quality food than baking as it destroys less of the amino acids
(especially lysine) than the higher temperature baking. However, it
is less conducive to large-scale production since much of its eating
qualities are associated with being freshly steamed. It loses quality
when re-steamed and its shelf life is short compared to baked bread
due to the higher moisture content. This will inevitably lead to what
the US and Europe call in-store baking as a means of bringing
large-scale industrialisation to the cities.
Indias second largest foodgrain crop is wheat, but strategically it
has tremendous and growing importance with an ever-larger popu-
lation, as it is a non-monsoon-based crop. It has six major growing
areas: the Northern Hill Zone (NHZ, 1.2 MHa), the North West Plain
Zone (NWPZ, 9.0 MHa), the North East Plain Zone (NEPZ, 9.0
MHa), Central Zone (CZ, 5.0 MHa), Peninsular Zone (PZ, 1.0 MHa)
and the Southern Hill Zone (SHZ, 0.2 MHa). Some 90% of Indian
wheat receives irrigation, although in the NHZ this does not occur at
higher elevations but closer to rivers unless the crops are close to a
river.
The NWPZ is a large fertile part of the Gangetic Plain and is more
than 90% irrigated, with crops maturing in 140 days and multiple
days with lows of less than 5 C. Wheat plants tiller well and develop
many spikes, so yields are high. However, disease can be a problem
due to mono-cropping (poor rotation). Temperature spikes at grain
fill can hurt yields in the same way as in the US Midwest. The NEPZ
187
is humid with a large number of minor rivers for irrigation, and it

also suffers from wheat diseases associated with a humid environ-
ment. Wheat matures here in 125 days and is susceptible to
unneeded rain showers at harvest time. The CZ and PZ are highland
deep soil areas but difficult to irrigate and may receive only two
applications of water per season, and with high temperatures we see
short growing seasons and poor yields.
India has developed and distributed 200 of its own varieties of
wheat since the 1980s which see several days with temperature lows
of less than 5C predominantly targeted for the higher yielding
regions. Disease control is effected by using the SHZ for plant
breeding. The many problems associated with Indian wheat produc-
tion have been solved around the world and will be in India as well,
although this will take time. These problems include but are not
limited to poor acceptance of new varieties and the widespread
planting of retained production from year to year, poor mechanisa-
tion, lack of modern harvesting methods and inexperienced machine
operators, which results in low-quality grain samples and lots of
admixture of foreign matter.
A considerable amount of Indian wheat is consumed as chapati, a
flat unleavened bread. The warm wheat areas have higher protein
than the cool NHZ. Hill wheats are widely used for biscuits/cookies.
PZ wheat is used for crackers and cookies due to its protein level and
quality. The baking industry is largely based in the south, which has
a deficit in wheat and the vast size of the country makes transport
expensive. Significant quantities are exported for hard currency, but
the industrialisation of baking and the introduction of whole grain
branded flour will lead to improvements in revenues. Better prices
for wheat will improve flour yields, and urbanisation will change
farming practises and consumption patterns, making India a signifi-
cant importer and producer of higher-quality breads for its
increasing population.
India will become an importer of higher-quality wheats over time,
which will have a significant impact on world wheat flows. With
only five classes (medium hard bread wheat, premium hard bread
wheat, biscuit wheat, durum and Khapli wheat, a particular Indian
wheat used for semolina) and no effective grading due to a largely
flat price structure for wheat within classes under the Indian state
run Public Supply Distribution (PSD) system, the Indian wheat
188
GRAINS AND OILSEEDS
market will evolve as it becomes more market driven. Most ethnic

breads (chapati, naan, tandori, rumati, roli, puri and bhatore) are
made from medium-hard bread wheat. The Indian government has
traditionally supported domestic wheat prices at a significant
premium to world prices and they have carried significant stocks to
allow it to intervene in the domestic food price.
Given the overall excitement in the wheat market of 2007, we will
review the run-up to this bull market, its causes and its effects. As
with all great bull markets, its roots lay in the previous years crop,
with widespread problems for the major exporters (US, EU-27,
Canada, Australia, Argentina and Russia). Since the early 2000s,
these countries have produced between 271 (200607) and 344 MMT
(200809), carried stocks as low as 36 MMT (end-2007) and as high as
73 MMT. Exports have ranged between 88 and almost 125 MMT and
stocks have responded dramatically, to build or draw-down, as price
signals have changed. At the beginning of 200607, their stocks stood
at 65 MMT and by the beginning of 200809 had fallen to 36 MMT, a
10-year low.
Of these countries, all but Russia has highly visible stocks. Russia
typically has a stock/use ratio of 10%. The EU has highly volatile
wheat production, producing 133 MMT in CY2002/03 and
111/147/132 MMT in the subsequent years, respectively. In CY2005
06, its crop decline year-on-year (yoy) of 14.5 MMT was absorbed by
the other major producers. The following year, however, saw disap-
pointing crops with the US down 8.0 MMT, the EU down 7.5 and
Australia down a disastrous 14.5 MMT. This 30.0 MMT dip was not
offset by the other major exporters and exports from the group
dropped to 88 MMT. Australia has a volatile, rain-dependent wheat
crop and, while production ranges between 10.0 and 30.0 MMT, it is
in fact rather binary, producing less than 15.0 MMT in dry years and
more than 25.0 MMT in wet years.
In 200607, the world became increasingly concerned with wheat
and the US drew stocks from 15 to 12 MMT, cutting domestic use and
exports. Similarly, the EU drew stocks by almost 10 MMT and also cut
domestic use and exports. Canada boosted exports and drew stocks,
and Australia halved its stocks to export more than 16 MMT versus
the previous years 23 MMT. Russia maintained big exports at 10
MMT, and Argentina almost emptied its stocks completely. At times
like these, we turn to the minor exporters (Ukraine, Kazakhstan, India,
189
China and Turkey) to see what they can contribute. The Ukraine crop
dipped from 18.0 to 14.0 MMT, and they drew stocks to maintain
exports at 3.5 MMT, down 3.0 on the previous year. Russia agitated for
a Ukraine export ban. Kazakhstan exports surged from 4.0 to 8.0 MMT
on a decent crop and a stock draw. India, however, who can be a 5.0
MMT exporter, was coming off two disappointing crops and had low
stocks. So, not only were they absent from the export market in
CY2006/07, but in fact imported almost 7.0 MMT. Chinas production
rose 11.0 MMT yoy, but they were already in stock-building mode and
withdrawing from the export market strategically. China therefore
barely exported 1.0 MMT more than the previous year. Turkey was
down to bare minimum stocks and had a sufficiently reduced crop in
CY2006/07 to be absent from the export market. In fact, across the
minor exporters there was a significant increase in imports yoy,
primarily lead by India and indicating the structural shift in the two
most populous countries in the world; China is now a structural
importer, and while India may come and go as both exporter and
importer, it will inevitably follow China to the structural importer
category.
Among the major importers, Egypt built stocks by 1 MMT in
2006/07 and increased imports yoy, Brazil increased imports by 1
MMT, Japan maintained imports, Indonesia raised imports and
Algeria cut theirs by an offsetting amount. South Korea, Nigeria,
the Philippines and Morocco cut imports modestly, while Iraq
imports took a big downturn and Mexico was unchanged. Overall,
major importer demand dipped by only 2.0 MMT in the face of a
30.0 MMT dip in major exporter production, pinpointing the very
staple nature of wheat demand and its insensitivity to price. It
should be clear to the reader that every large market player has
access to the shipping fixtures, or grain movements, by loadport
and discharge port.
We then entered the major bull run. Any problems in the 2007
growing season would cause a major disruption, and the hedging
pressure and speculative pressure increased to intense levels. US
production rebounded by 6.5 MMT in CY2007/08 and another 12.0
MMT in CY2008/09, but only after drawing stocks to a low 8.0 MMT.
Disastrously, the EU had more problems in 2007/08 and production
dipped another 5.0 MMT, and stocks hit a near record low. By
CY2008/09, a world-saving rebound of 31 MMT would be
190
Table 7.7 Rice 5-,10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 199092). After
each country name the % of world area devoted to this crop in 2012 and 1992 are given
Rice Last three-year average (201012) 200507 200002 199597 199092 % of world
Brazil 2% 3% 2.5 3.250 8 12% 18% 5% 18% 44% 18% 27% 77% 30% 41% 106% 22% 113%
Thailand 7% 6% 11.0 1.875 20 4% 5% 9% 7% 10% 18% 15% 22% 41% 21% 34% 61% 65%
China 19% 22% 30.0 4.750 141 4% 6% 10% 5% 6% 12% 4% 8% 4% 8% 17% 8% 165%
India 27% 29% 43.5 2.250 100 1% 7% 6% 0% 20% 20% 1% 22% 23% 2% 32% 35% 78%
World total 100% 100% 158.5 2.875 461 3% 6% 9% 6% 11% 18% 6% 15% 21% 8% 21% 30% 100%
Source: Adapted from Informa
GRAINS AND OILSEEDS

191
harvested, but not until the market had gyrated wildly. Adding to
the woes in 2007, Canadian wheat production dipped 5.0 MMT, and
they too drew stocks heavily to record a low of barely 4.0 MMT.
Australian production recovered, by a mere 2.8 MMT, to a sub-14.0
MMT crop. Argentinian and Russian crop production increased
slightly, and the major exporters saw their total production increase
by 6 MMT and stocks draw another 7.0 MMT on top of the previous
years 20.0 MMT decline. Collectively, their production would surge
by more than 66.0 MMT in CY200809 to end the bull market. Minor
exporters had a domestic production rebound of 8.0 MMT but
reduced their exports yoy, and while they cut their imports in half
they were also building stocks. Although there was some variance
between major importers, stock were built modestly and imports
rose modestly.
Wheat exhibited the classic volatility of a market with inelastic
demand and whose price-solving mechanism is to scale a steep
marginal supply curve to increase production at the expense of
competing crops. This occurred at the same time as crude oil price
was increasing dramatically and maize demand for ethanol surged
in the US. As in Table 7.6, the wheat supply from 200507 to 201012
would only increase in area by 2%, yield would rise 7% and produc-
tion by 10%. Production increases were 20% in the EU, 18% in China,
14% in CIS/FSU, 13% in India and 9% in the US.
As a foodgrain, rice provides the most widely consumed staple
food of over half the worlds population (see Table 7.7), especially in
Asia and the West Indies. It is the seed of the monocot plants Oryza
sativa (Asian rice) or Oryza glaberrima (African rice). It is the predom-
inant dietary energy source for 17 countries in Asia and the Pacific,
nine countries in North and South America and eight countries in
Africa.
Rice provides 20% of the worlds dietary energy supply, while
wheat supplies 19% and maize 5%. It is the grain with the second-
highest worldwide production after maize, but since a large portion
of maize crops are grown for purposes other than human consump-
tion, rice is the most important grain for human nutrition and caloric
intake, providing more than one fifth of the calories consumed
worldwide by the human species. There are many varieties of rice
and culinary preferences vary regionally. In the Far East, there is a
preference for softer and stickier varieties.
192
GRAINS AND OILSEEDS
Rice yields continue to grow while area is largely stagnating,

Chinas dramatically better yields than India means it produces 40%
more rice from two-thirds as much land. Indias yield growth at 33%
in 20 years is, however, twice that of China, and Brazils yield has
doubled in the same period. As previously mentioned, the Indian
crop is monsoon-driven. World trade is small and most countries
that consume rice grow their own.
To close the foodgrain section, some basic numbers are provided
for processed food sales. Worldwide, they are approximately
US$3.5trillion, and the industry is growing. The processors are giant
companies that own huge brands, the CPGs such as Nestl SA. These
companies dwarf the ag majors who are their suppliers. The food
industry is a complex, global collective of diverse businesses that
supply much of the food energy consumed by the worlds popula-
tion. Onlysubsistence farmers, those who survive on what they
grow themselves, can be considered outside of the scope of the
modern food industry.
In developing country markets, the two reference points are the
US and the UK. With populations of 313 million and 55 million,
respectively, they can be used to estimate what the food economies
of less-developed countries will likely look (more) like in the next
few years.
In the US, consumers spend approximately US$1.3 trillion annu-

ood-expenditures.
million people are
consumer base of
example, has the
UK grocery market
GDP, and employs
uring sector in the
K manufacturing.
ufacturing in the
. This is roughly a
, for example, has
193
194

Table 7.8(a) Soybean 5-,10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 1990
92). After each country name the % of world area devoted to this crop in 2012 and 1992 are given.
Soybeans Last three-year average (201012) 200507 200002 199597 199092 % of world
US 29% 42% 30.5 2.750 85 8% 2% 5% 4% 8% 12% 17% 11% 30% 31% 18% 54% 110%
Argentina 18% 9% 18.5 2.625 49 16% 7% 9% 62% 2% 60% 191% 20% 248% 286% 12% 333% 105%
Brazil 24% 18% 25.5 2.875 74 18% 7% 27% 57% 6% 66% 114% 25% 166% 154% 53% 288% 115%
Paraguay 3% 2% 3.0 2.125 6 19% 8% 28% 105% 18% 68% 155% 6% 139% 222% 39% 348% 85%
China 8% 14% 8.0 1.750 14 15% 10% 6% 14% 4% 11% 1% 3% 2% 8% 26% 36% 70%
India 10% 5% 10.0 1.125 11 22% 10% 34% 73% 27% 120% 96% 18% 132% 223% 21% 294% 45%
World Total 100% 100% 105.0 2.500 258 13% 1% 15% 33% 5% 39% 63% 15% 87% 89% 25% 135% 100%
Table 7.8(b) Rapeseed 5-,10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 1990
Rapeseed Last three-year average (201012) 200507 200002 199597 199092 % of world
Canada 22% 15% 7.5 1.750 14 38% 5% 45% 88% 32% 143% 69% 34% 127% 164% 37% 258% 100%
China 20% 31% 7.0 1.750 13 15% 1% 14% 2% 16% 18% 10% 32% 45% 23% 45% 78% 100%
India 20% 31% 7.0 1.000 7 6% 5% 11% 44% 15% 65% 4% 9% 14% 12% 13% 27% 57%
EU 27 19% 15% 6.5 3.000 20 17% 1% 17% 57% 1% 56% 58% 94% 186% 117% 11% 145% 171%
World total 100% 100% 34.0 1.750 61 25% 2% 28% 47% 16% 70% 47% 28% 87% 75% 31% 131% 100%
Table 7.8(c) Sunseed 5-,10-, 15- and 20-year AYP progression (comparing 201012 with 200507, 200002, 199597 and 1990
Sunseed Last three-year average (201012) 200507 200002 199597 199092 % of world
US 2% 6% 0.5 1.625 1 19% 2% 19% 30% 5% 35% 40% 8% 35% 22% 11% 14% 108%
Argentina 6% 15% 1.5 2.125 4 27% 25% 9% 16% 23% 3% 44% 18% 34% 29% 38% 2% 142%
EU 27 16% 24% 4.0 1.875 7 12% 15% 28% 15% 26% 45% 31% 96% 48% 3% 37% 66% 125%
CIS/FSU 53% 26% 13.0 1.375 18 32% 15% 51% 84% 42% 160% 106% 43% 195% 179% 7% 199% 92%
World total 100% 100% 24.5 1.500 37 9% 16% 26% 23% 28% 57% 23% 23% 52% 44% 16% 68% 100%
Source: Adapted from Informa
GRAINS AND OILSEEDS

195
OILSEEDS VEGETABLE PROTEINS

We harvest 355 MMT of the three major softseeds (soybeans, rape-
seed and sunseed, (see Tables 7.8 a, b and c)), of which 260 are
soybeans, 60 are rapeseed and 35 are sunseed. The major soybean
economies are the US (85), Argentina (50), Brazil (75) and Paraguay
(5). For rapeseed the big four economies are Canada (15), EU-27 (20),
China (15) and India (5). For sunseed the big three economies are
Canada (5), EU-27 (5) and CIS/FSU (20). The softseeds yield
vegetable oil and high-protein meal in ratios of 19/35/33% of oil and
79/63/65% meal from soy/sun/rape crushing, the balance being the
hulls or shells. This means a global soft oil output of roughly 50, 12
and 20 MMT, respectively. Much of the softseeds are crushed and
their products consumed locally. Historically, we have described
China, or Asia, as oil deficit and the EU as meal or protein
deficit. Global exports of the three softseed oils are estimated by
Agrimax at 8.0, 5.0 and 3.0 MMT respectively, compared to global
palm oil flows of more than 38.0 MMT.
In the early 1990s, the US dominated global soybean production.
By a decade later, Brazil and Argentina combined produced as much
as the US, and by the early 2010s Brazil alone threatened to match the
US in production. World planted area has grown by almost 90%.
World yields have grown by 25%, and production has surged by
135%. Brazil has the highest yields, followed by the US and
Argentina, but it is important to note that yield advancements are in
decline and largely occurred during the 1990s.
In contrast, hard oils (their physical state at room temperature) are
produced primarily from fruit (as compared biologically to seeds).
The most commercial is palm oil, but the family also includes
coconut oil and others. Butter and lard (animal fat) are also included
in this category. Malaysia and Indonesia dominate palm oil produc-
tion and annually export some 19.0 MMT each, amounting to more
than 65% of global vegoil (the common industry abbreviation for
vegetable oils) trade flows. Since they come from fruit, there is an
associated pulp that remains from processing, normally returned to
the soil as fertiliser.
The four countries that dominate world rapeseed production,
with 90% of production, are all major wheat producers, and rapeseed
grows very nicely in a rotation with wheat. EU yields are more than
170% of the world average and planted area is now more or less
196
GRAINS AND OILSEEDS
equal to each of the other producing countries, at 7 MHa each. EU

area planted has driven the global yield growth, expansion being
primarily for biodiesel production. Indian yields are abysmal, but
domestic consumption is protected by the large costs of importing
rapeseed from major producers and trucking to the interior. China
dominates global rapeseed trade.
Sunseed production is driven by three countries (CIS, EU and
Argentina 49%, 19% and 11%, respectively), with the US a poor
fourth at 3%. While production has tripled in 20 years the low yields
and expanding area in the CIS has slowed the market growth. Yields
have expanded healthily in any case, although there was a lost
decade when CIS yields regressed rather than grew. Sunflower oil is
sold at a premium compared to other vegetable oils into to
Mediterranean and North African markets, where it is preferred as a
cooking oil. Global production at less than 40.0 MMT will grow to
50.0 or 60.0 MMT.
Each major vegetable protein economy produces a variety of
proteins locally as determined by their comparative advantage, and
the balance is either exported or imported. Since China is a major
importing vegetable protein economy, its S&D balance is
summarised here. Its domestic production of almost 60.0 MMT of
major oilseeds is dominated by soybeans (15), sunseeds (2.5) and
rapeseeds (12.5), as well as cottonseeds (12.5) and groundnuts (15),
the non-US name for peanuts. In addition, it imports a staggering
60.0 MMT of soybeans and 2.5 MMT of rapeseeds, crushing 100.0
MMT per annum. In the early 2000s, China imported only 20.0 MMT
of soybeans and had a major softseed crush capacity of 55.0 MMT, of
which 25.0 MMT was soybeans. At that time, the big four soybean
exporters (US, Argentina, Brazil and Paraguay) had a total crush
capacity of 100 MMT which has grown over the same 10-year period
to roughly 125 MMT, (US 45, Argentina 38, Brazil 40 and Paraguay 3)
while Chinas grew from 25 to 100 MMT.
Softseed crushing is the process by which seeds are pressed
through a die. Heat, steam and solvents are used to extract the oil
from residual meal to form the two principal by-products and leave
the hulls and seed covering. One can imagine other softseeds that are
produced and used in different ways, such as peanuts, sesame seeds
and mustard seeds, which are consumed and processed or simply
cooked. Oilworld.biz, an analyst specialising in the vegetable oils
197
business, comprehensively analyses the crushing of 10 major

oilseeds, 17 major oils and 12 major meals, including cottonseed oil,
fish oil, corn oil, palm and palmkernel oil, butter and lard. Local
tastes and GDDs drive the local markets, and deficits and surpluses
are imported or exported. Rapeseed oil is preferred for its taste in
Chinese cooking and it is by far the largest rapeseed market in the
world.
The complexity of the soybean business is perhaps best illustrated
by the soybean product tree (see Bell, David E., and Mary L.
Shelman, 2006, Bunge: Poised for Growth, Harvard Business
School Case 506036, July), which shows that the crushing plant
complex is quite large and comparable to a petroleum refinery if all
the various streams are included. In North America and the EU, the
crushing plant will supply downstream processors such as Solae (a
Bunge DuPont joint venture) for further processing. In Brazil, the
industry is still evolving to develop the various processed product
streams and many of these crushing plants are truly biomass opera-
tions for instance, the plant being built on 10,000 Ha of which 20 Ha
is the actual plant, bottling and bagging, trucking, warehousing and
logistics, and the balance is producing eucalyptus trees which are
harvested and used to power the plant and its various services.
What we should note at this point is these plants do not suddenly
appear, they are planned in advance and the crushing equipment is
ordered in advance. The storage facilities for vegoil are quite tech-
nical, and meals are not without their complexities due to their
physical characteristics. Fundamental analysis includes the fore-
casted change in crush and downstream capacity by location and
type of operation. Compared to petroleum refineries, they are cheap.
Impressive worldscale operations are built for US$0.2 billion rather
than US$2.0 billion. Within the study mentioned above, at that time
Bunge was the worlds biggest soybean crusher, and the expectation
for soybean crush evolution by geography is given. They expected
the 2010 crush of soybeans to look like US/Arg/Braz/EU/China as
52/25/30/17/25 (149 MT total), respectively, when in fact it looked
more like 45/38/40/12/76 (211 MT total), respectively. The forecast
missed both the size and geography of actual growth. The error was
7/13/10/5/51 (62) or 87/152/133/71/304 (142)% of forecast.
Although these would have been constantly revised by Bunge, it
demonstrates the ease of making a substantial error and consequent
difficulty of building for the future. In fact, the marketplace did
198
GRAINS AND OILSEEDS
generally underestimate Chinas appetite for soybean imports and

the desire to crush them domestically, as it did the Chinese growth
story for all commodities.
We also have to remember that once capacity is built it is normally
run, in any industry, at any contribution to the bottom line that
exceeds variable cost. The different regional growth patterns imply
different rates of port development to move the commodities, as well
as different rates of growth in downstream and associated indus-
tries. A further unexpected agribusiness consequence, but one which
the petroleum business is quite familiar with, was the building of
strategic reserves of foreign currency held in commodities by China
look at their enormous stocks of soybeans and oilseed rape, 16 and
6 MMT, respectively.
The Chinese domestic oilseed complex growth has been virtually
stagnant since the early 2000s growing from 55 to more than 58
MMT, within which soybeans contracted by 2 MMT, while rapeseed,
cottonseed and groundnuts increased. Soybean imports grew from
21 to almost 60 MMT, and rapeseed imports grew by more than 2.5
MMT, so that crush now stands at soy 61 MMT, rapeseed 15 MMT,
sunflower 1 MMT, cottonseed 10 MMT and groundnuts 7 MMT, for
a total crush of 96.0 MMT the biggest in the world. Add to that the
24 MMT of oilseeds consumed other than through full crush (partial
processing), of which 11 MMT is soybeans, 3 MMT is cottonseed and
8 MMT is groundnuts, and we see a better picture of the 120 MMT
Chinese oilseed powerhouse, consuming far more than any of the
big three producers. In the early 2000s, they carried oilseed stocks of
almost 19 MMT in China, and in 2013 stocks stand at an estimated 24
MMT. The USDA estimates that if it costs US$100/MT to move
soybeans from Iowa to Chinese ports, it costs US$175 from Mato
Grosso in Brazil. This means the expansion of soybeans in Brazil is
disadvantaged at the farmgate by that amount, and this inevitably
leads to Brazil finding other uses for soybeans until transport effi-
ciencies can be de-bottlenecked.
On the soybean demand side, 10 years is also a long time and the
rapidly changing face of Brazilian agribusiness is well illustrated by
the emergence of JBS on the world stage. JBS is the largest Brazilian
multinational food processing company, producing fresh, chilled
and processed beef, chicken and pork, and also selling by-products
from the processing of these meats. This has lead to a surge in
Brazilian soybean consumption domestically. In a decade, its
199
domestic meal consumption has grown by 5 MMT to 12.5 MMT.

Argentine meal consumption in the same period grew from 0.25
MMT to almost 2.0 MMT.
JBS has established itself as the largest global company in the beef
sector with the acquisition of several retail chains and food compa-
nies in Brazil and around the world, especially the 2007 US$225
million acquisition of US firm Swift & Company, the third-largest US
beef and pork processor, renamed as JBS USA. It leads the world in
slaughter capacity, at more than 50,000 head per day, and continues
to focus on production operations, processing and export plants,
nationally and internationally. With the new acquisition, JBS entered
the pork market, featuring an impressive performance in this
segment, to end the year as the third largest producer and processor
of this type of meat in the US. The acquisition expanded the
companys portfolio to include the rights for worldwide usage of the
Swift brand. The following year, JBS acquired Smithfield Foods beef
business, which was renamed JBS Packerland. JBSs production
structure is embedded in consumer markets worldwide, with plants
installed in the worlds four leading beef producing nations Brazil,
Argentina, US and Australia serving 110 countries through
exports. In September 2009, JBS announced that it had acquired the
food operation of Grupo Bertin, one of three Brazilian market
leaders, consolidating its position as the largest beef producer in the
world. On the same day, it was announced that the company had
acquired 64% of Pilgrims Pride for a bid of US$800 million, estab-
lishing JBSs position in the chicken production industry. In August
2010, it was reported that JBS was trying to sell some of the eight
slaughterhouses it owns in Argentina because of scarce livestock
and export restrictions. By 2011 they were attempting to gain
control of Sara Lee Corporations meat business.
Brazil has aquaculture production targets of 1.0 MMT by 2015 and
10.0 MMT by 2020 from a base of 0.5 MMT in 2011. While this may be
too high to achieve by 2020, we can easily imagine them managing it
by 2025, again reshaping soya and (non-vegetable) protein flows. It is
worth looking back at Table 7.5 to understand the significance of this
number.
In 2003, the unthinkable happened in the soya world: there were
terrible crop yields in the US, Argentina and Brazil, all in the same
year. Forecasters expected a rising yield, but the US dipped from 2.56
200
GRAINS AND OILSEEDS
to 2.28 MT/Ha, Argentina from 2.82 to 2.36 and Brazil from 2.82 to
2.37. In 2002, Argentina and Brazil out-yielded the US for the first
time, and their combined production matched US. For the first time,
not only did global soybean production not grow, it dipped by some
10 MMT. This sparked an unprecedented rally which had long-term
effects on how the markets traded, who dominated them and how
the Chinese thought about their soybean strategy. In the authors
opinion, this drop may be partly attributed to the illegal spread of
GM seeds in Latin America at a time when the technology was new
and certainly undeveloped for Latin American conditions. Critically,
it demonstrated that yield advancements came with increasing yield
variability and unexpectedly large sensitivity to weather variations.
The US saw 0.5 standard deviation changes in GDDs give far bigger
swings in yield than history would have lead us to expect.
For those who follow freight markets, part of Chinas soybean
importing strategy has been to add Chinese tonnage to the global dry
bulk market, since they are structurally short, causing a sharp down-
ward correction in freight prices.
HOW ROTATION CONVERGES THE GRAINS

As a major source of income for trading companies and hedge funds
alike (see Appendix 7.1), and definable by excellent fundamental
analysis, we can arbitrage maize, wheat and soybean prices. In the
short run, one can reasonably expect these three commodities to
change price relative to each other, to reallocate or switch hectares
between crops and hemispheres. We can always bring more land
into production, but in Brazil, for example, that involves a year of
land clearance of indigenous plants before a year of growing rice and
clearing the land, and then a serious commercial crop can be started
in the third year. Table 7.9 shows an interesting view of the major
crop economies in a side-by-side comparison of total arable land flex-
ibility and individual crop flexibility. The major opportunities with
existing resources, in terms of area, are all within Table 7.9. The
serious student should understand this one table representation of
flexibility in both percentages and individual crops as well as the
yield gaps presented in the various tables for the major crops, by
country.
The US and the EU-27 are the most economically responsive areas
or rational actors to relative price, by which we mean per-hectare
201
Table 7.9 Rotation flexibility by major grain economy in 20 years (total arable area and min/max percentages by major crop)
Swingable hectares is total area times (max minus min); current percentage devoted to each crop is given by country and world
Area Maize Barley Sorghum Wheat Rice Soybeans Rapeseed Sunseed
MHa Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max
US 90 32% 41% 2% 6% 22% 34% 28% 38%

EU-27 60 9% 17% 21% 34% 40% 46% 6% 12% 6% 15%
Argentina 30 9% 17% 2% 5% 11% 38% 31% 68% 6% 19%
Brazil 45 28% 46% 3% 11% 5% 15% 32% 59%
China 105 22% 33% 23% 31% 29% 34% 7% 10% 6% 8%
India 90 31% 34% 48% 57% 3% 12% 6% 9%

CIS/FSU 85 2% 9% 16% 38% 53% 64% 5% 17%
Swingable hectares
US 7.9 0.0 3.5 10.9 0.0 8.8 0.0 0.0
EU-27 4.8 8.1 0.0 3.8 0.0 0.0 3.8 5.2
Argentina 2.2 0.0 0.9 7.9 0.0 11.3 0.0 4.0
Brazil 8.3 0.0 0.0 3.2 4.5 12.4 0.0 0.0
China 12.4 0.0 0.0 8.5 4.7 3.6 2.7 0.0
India 0.0 0.0 0.0 2.7 8.7 7.6 2.7 0.0
CIS/FSU 5.3 18.6 0.0 9.6 0.0 0.0 0.0 9.6

Theoretical total 41 27 4 47 18 44 9 19
Current % Maize Barley Sorghum Wheat Rice Soybeans Rapeseed Sunseed

US 40% 0% 2% 22% 0% 35% 0% 1%
EU-27 16% 22% 0% 44% 0% 0% 11% 7%
Argentina 12% 0% 4% 11% 0% 66% 0% 6%
Brazil 31% 0% 0% 4% 5% 59% 0% 0%
China 33% 0% 0% 23% 29% 7% 7% 0%
India 0% 0% 0% 33% 48% 12% 8% 0%
CIS/FSU 9% 17% 0% 59% 0% 0% 0% 15%
Current % world 21% 6% 5% 28% 20% 13% 4% 3%
202
Source: Agrimax.
GRAINS AND OILSEEDS
income. Since the early 1990s the US has planted as little as 32% and
as much as 41% of its 90.0 MHa of arable land to maize, 26% to
sorghum, 2234% to wheat and 2838% to soybeans. At the last
count, the US were at maximum on maize, 35% on soybeans and
minimum on wheat. This trend will continue with more ethanol
(maize) produced and less land available for wheat and soybeans.
Wheat area is the most switchable, and surged 3.0 MHa in 2003.
Soybean hectares surged almost 2.5 MHa in 1997 in response to the
Freedom to Farm Act. Over the 20 years, total land area only
increased by 6 MHa. With the threat (or reality) of E15, it is expected
there will be more maize at the expense of wheat.
By contrast, the EU-27 has 60.0 MHa in grains and oilseeds up by
almost 20.0 MHa's in 20 years, with maize swinging between 9% and
17%, wheat between 40% and 46%, barley between 21% and 34%,
rapeseed between 6% and 12% and sunseed between 6% and 14%. At
the last count, the EU-27 was close to maximum on wheat and rape-
seed, average on sunseed and close to bottom on barley.
Argentina and Brazil till some 30 MHa and 46 MHa, respectively,
with each having grown from 15.5 and 30.5 since the early 1990s.
Argentina is more rotationally complex, with 1017% maize, 25%
sorghum, 1138% wheat, 3168% soybeans and 619% sunseed.
Brazil is 2846% maize, 311% wheat, 515% rice and 3259%
soybeans. Latterly, Argentina has been in the middle on maize, at the
high end for sorghum, at the bottom end for wheat and all the way to
max on soybeans and at minimum for sunseed. Brazil was close to
minimum for maize, bottom end for wheat and rice and, like
Argentina, at max for soybeans.
China, with 103 MHa under tillage, is almost unchanged in area
since the early 1990s (+5 MHa), and can swing 2233% on maize, 23
32% on wheat, 2934% on rice, 711% on soybeans and 58% on
rapeseed. At the last count, it was max on maize (to blend with
imported soybeans), minimum on wheat, rice and soybeans and
close to max on rapeseed. The main China growth story is meat
production pork and chicken with high FCE. A high FCE requires
a singular focus on maize-plus-soymeal diets, for physical flowa-
bility or product handling as well as nutrition.
India, with more than 90 MHa in tillage, swings only 3134%
wheat, 4857% rice, 412% soybeans and 69% rapeseed. Food secu-
rity points to more wheat over time but much of this is going to go to
203
more intensive large-scale farming. Indian productivity per hectare

has only one way to go: up.
The CIS/FSU has some 85 MHa under tillage and dismal yields.
Maize farming should be declining and swings 29% (currently at
max), wheat 5364% (currently in the middle), 1638% barley
(currently at the low end) and 517% sunseed, which is now at the
high end. We expect to minimise barley and maximise sunseed and
wheat for the foreseeable future.
It would be nice to make a big deal of Argentina and Australia, but
this is not realistic. They do not have the land mass or yields and so,
even if Canada is max on rapeseed at 1.0 MHa, it simply does not
make a global difference. At this time, it is max on rapeseed and
minimum on wheat. Australia is dryland farming with sporadic rain,
so unreliable. The call-like planting of Australian wheat means
they will continuously plant, from year to year, and hope for rain just
as Texas does in the US.
Area times yield equals production. The most populous countries
have the land pretty much tapped and China has done tremendous
work on yield. The baton falls to India to improve crop husbandry.
Brazil has land in abundance but infrastructure is so tight and expen-
sive that it is likely to continue its domestic trend toward more meat
and aquaculture production. This would expand its export capacity
by displacement, just as it now moves vast quantities of sugar by
container to the export market. The major opportunities with existing
resources in terms of A are all within Table 7.9 and the serious student
should understand this one table representation of flexibility in both
percentages and individual crops as well as the yield gaps presented
in the various tables for the major crops, by country.
SUMMARY OF MAJOR TRENDS AND SWING FACTORS FOR

THE FUTURE
If one thing alone has changed the grain markets completely since
the early 1990s and is likely to continue to do so, it is undoubtedly the
US corn-based ethanol programme. It remains phenomenally diffi-
cult to change commercial US law once it is in place other than by
incremental amounts. If cellulosic ethanol arrives it will change the
world forever and cause grain prices to collapse. However, it
appears to be no closer in terms of substantial economic reality than
we saw in the early 2000s.
204
GRAINS AND OILSEEDS
If there is a second thing that has also changed the grain markets
completely during this period, it is the manner and rate at which
CPGs are growing to dominate our increasingly urbanised food
consumption. At the time of writing, Chinas Shuanghui
International has just bought Smithfield Foods, the huge US-based
but globally active pork and meat company, for US$4.7 billion. The
need for modern food processing safety, branding and packaging,
and all the required supply chain management skills, has rendered it
more cost effective to buy it rather than build it.
If there is a third thing that must happen over the next few years, it
is the intensification of agriculture for the cost of bringing more
area into production has become much more expensive than most
had anticipated.
Since maize combined with soybean meal is the cornerstone of
modern animal (and soymeal for aquaculture) nutrition, much more
grain will be consumed in Brazil and exported as meat. China and
the US have some 34 MHa under maize, and both will increase area.
Also, Chinese yield will move towards the US (there is a 3 MT/Ha
gap, see Table 7.3), just as China did with the EU in wheat (see Table
7.6). The maize market into the 2020s will remain fundamentally
tight and expensive. E15 will take more corn to the fuel tank,
although there are some real costs being discussed at the retail petrol
station level where the retail supplier is pushing hard to stay at E10
or go to E15, but not carry both. This would require adding pumps,
tanks, trucks and re-branding all expensive items. Brazil will export
more maize than the US consistently. The only two things that can
cause maize demand to break to the downside are a dramatic u-turn
in US energy policy (1:100) or a breakthrough in cellulosic ethanol
(1:50). Even a dramatic fall in crude oil prices would only stimulate
maize demand for the gasoline pool as it worsens the economics for
cellulosic ethanol. Economics says Brazilian ethanol should continue
to flow in ever-greater quantities to the US, but it may not become a
political reality.
It is ironic that the CIS/FSU has a higher barley than wheat yield,
something almost impossible in terms of modern farming. The
CIS/FSU has the greatest potential to increase yield through intensi-
fication and plant breeding, and has some 49 and 14 MHa under
wheat and barley, respectively. Any area reductions will be offset by
increased commercialism of these two markets inside Russia, from
205
farmgate to consumer. Wheat will continue to assume the role of

primary determinant of grain prices globally as its volume is
increasing while US maize volumes decline, net of ethanol. India will
become a consistent importer of wheat and withdraw from the
export market into the early 2020s.
The Chinese (Asian) and Indian (sub-continent) appetites for soya
will continue as meat and fish demand increase. The CPG intensifica-
tion of their food systems will also increase, along with urbanisation
and wealth. At 11 MMT of soybeans and 7 MMT of rapeseed, soft-
seed production in India is growing rapidly, and significant imports
will come in time.
We have been waiting for palm oil production to reach a
maximum in Malaysia and Indonesia, but it continues to increase. At
some point this must happen and will create more pressure for
global soybean area to increase.
In terms of AYP, we will continue to see area expand slowly but
yield to expand at more impressive rates (see Table 7.1). In fact, the
author is optimistic it will be much higher.
APPENDIX 7.1: AGRIBUSINESS INVESTORS

The ag investing funds are listed below.
Commodity-specialist funds: Ospraie, Ospraie Wingspan,

Touradji, BlackRiver, Armajaro, etc;
Global Macro funds: DE Shaw, Soros, etc;
Pension funds: APG, Calpers, BT, Hermes, TIAA-CREF, etc;
Sovereign Wealth funds (all EM-based and EM in focus): Kuwait
Investment Authority, National Bank of Dubai, SinoLatin
Capital, etc;
Private Wealth aggregators: Barclays Global Investors (now
BlackRock), GSAM, Adecoagro (Soros), etc;
Index funds: the GSCI, DJ-AIG etc index funds and their
hedgers, etc, as well as Schroders in ags;
The Mega funds (ABC): Ashmore, Blackrock, Carlyle, etc;
Managed Futures industry: self explanatory;
Endowment funds: Harvard, etc; and
Private Equity: BlackRiver (Cargill), The Mega Funds, Louis
Dreyfus (Calyx Agro) and the L-D family, as well as PAI and an
endless list stretching to CP (Charoen Pokphand) and Glencore.
206
8
Coal
Jay Gottlieb
This chapter will provide the risk management professional with an

orientation to understand the oldest and oddest of energy markets:
coal. It will explain the physical characteristics of coals, coal market
structure and dynamics, and the coal price indexes and trading
venues used for transacting financial derivatives. The chapter will
also cover key developments in the fundamentals of coal along with
an understanding of the broad range of instruments available to
manage risk in that market, and will provide an overview of market
drivers and their interaction, as well as offer an initial reference for
the detailed data needed to analyse the coal market.
OVERVIEW
Coal seems to be the unwanted stepchild of the energy world: dirty,
old-fashioned, not really popular anymore. Who cares? On the other
hand, those who do care a lot often seem to echo the famous words of
a White House adviser on energy and the environment:
A Harvard University geochemist who serves as a scientific adviser
to President Obama is urging the administration to wage a war on
coal.
The one thing the president really needs to do now is to begin the
process of shutting down the conventional coal plants, Daniel P.
Schrag, a member of the Presidents Council of Advisers on Science
and Technology, told the New York Times. Politically, the White
House is hesitant to say theyre having a war on coal. On the other
hand, a war on coal is exactly whats needed.1
Trends in worldwide coal consumption indicate that Professor

Schrags war is going badly:
207
Coal consumption grew by 2.5% in 2012, well below the 10-year

average of 4.4% but still the fastest-growing fossil fuel...Coal reached
the highest share of global primary energy consumption (29.9%)
since 1970.2
Coal has driven global development since the British industrial revo-
lution, beginning in the 18th Century with the harnessing of
increasing amounts of coal-fired steam power for transportation and
steel production. The role of coal-fired steam in transportation and
manufacturing along with the use of coking coals in the production
of steel is familiar. While other fossils remain a big part of peoples
daily lives petrol for cars and natural gas for home heating and
cooking coal has largely receded from view. It works away quietly
in the industrial background. While coal is no longer used locally for
transportation or building heat, it is still consumed as a key compo-
nent in steel and cement production and fuels around 40% of the
worlds electric power generation.
Coal is found abundantly around the world, is relatively easy to
produce with existing mining technologies and can be transported
through a wide variety of modes, such as conveyor belt directly from
mine to power plant, or through combinations of truck, rail, barge
and ocean-going freighter. As transportation infrastructure devel-
oped around the world since the 1960s, prices for bulk transportation
declined and coal changed from a commodity with only a local
regional reach and economics to one that is traded similarly to other
higher-value energy commodities, flowing around the world from
production areas to wherever it commands the highest value in
consumption. Along with the explosion of transportation options,
coal consumers have become much more sophisticated in managing
their power plants to run on a greater variety of coals, adjusting for
physical and chemical differences in coals from divergent sources.
The major exporters of coal are Indonesia, Australia, South Africa,
Colombia, US and Russia. China and Europe are the major
importers. While the exact numbers will of course change from year
to year, the major participants will not.
CHARACTERISTICS OF COAL
So, let us return to the question, what is coal? It is an energy-rich
source of carbon that is relatively easy to find, mine and transport,
but is also bulky and heavy relative to its energy value. Also, coal
208
COAL
comes with many other non-carbon components that must be

controlled to limit pollution and other unwanted emissions from
power production and other consumption.
Coal is a combustible sedimentary organic rock consisting of more
than 50% carbon by weight. It is a fossil fuel derived from plants that
grew in swamps that were later buried by sediments. Geological
processes compressed and heated the plant remains over vast
periods of time, producing various ranks (or categories) of coal.
With increasing rank, coal becomes harder, brighter and the heat
content is higher. The ranks from lowest to highest are: peat,
brown coal, lignite, sub-bituminous, bituminous and anthracite
(listed in Table 8.1).
While coal is chiefly comprised of carbon, hydrogen and oxygen,
it also contains varying amounts of sulphur, nitrogen and other
elements. Coal quality varies a great deal and is priced based on
these characteristics. The heat content is the key value of the
commodity for electricity generators and cement producers, while
other characteristics are important for steel producers. Disposing of
the non-desirable components, particular sulphur and nitrogen,
adds cost to the consumption of coal.
Table 8.1 Coal rank description
Peat Wet plant material that has been subject to bacterial

and fungal action, very low energy level, moisture
level ~60% calorific value ~2,600 kcal/kg
Brown coal Peat that has had the water squeezed out, plant
remains still visible moisture ~50%, calorific value
2,800 kcal/kg
Lignite Coal is hard and massive, black looking, moisture

content 4050%, calorific value about 4,000 kcal/kg
Sub-bituminous Coal is hard, brittle, black and shiny, moisture content
is 2040%, calorific value 4,0005,800 kcal/kg
Bituminous Coal is softer and shiny, moisture content is 820%,
calorific value is 5,8008,000 kcal/kg, crucible
swelling number from 29+ possible for coking coals,
volatile matter 1640%
Anthracite Coal is very shiny, repels moisture, calorific value
7,8008,000 kcal/kg, no coking properties
209
Heat content is measured as the heat produced by combustion of a

specified quantity of the fuel when burned at a constant pressure
under controlled conditions for water vapour. It is measured in
terms of either British thermal unit (Btu) per pound in the US or kilo-
calorie per kilogram (kcal/kg) internationally. In all cases, higher
heat content is preferable to lower.
Thermal coal fires power generation plants, and metallurgical (or
met) coal is used for steel production. We will now look at which
coal qualities are of importance in thermal and metallurgical
consumption.
Thermal coal
Most coal is used for the energy content within the volatile matter
and the fixed carbon. These coals are generically termed thermal
(or steam) coals and are mostly used for electricity generation. A
typical Australian thermal coal contains 6,080 kcal/kg of usable
energy (net as-received energy) or 25.46 megajoules/kilogram
(MJ/kg) of coal. Electrical energy (power) is measured in watts
which are joules per second, therefore one kilowatt hour of electricity
(one unit) converted from coal at 35% efficiency requires 10.286 MJ of
coal energy every hour, or 0.404 kg of coal. Other thermal coal uses
are the calcination (breakdown by heat) of limestone to form cement
for construction industries or lime for agricultural purposes.
Hospitals and other institutions use coal for process heat, as do abat-
toirs, wool sours and timber-drying processes.
Metallurgical coal
For steel and other metallurgical production, certain bituminous
coals are particularly suited to release gaseous components, called
volatile matter, when heated to extremely high temperatures in the
absence of oxygen. When these special bituminous coals swell on
heating above 350 0C and release their volatile matter, they leave
behind a hard porous carbon residue called coke. These coals are
called coking coals and are limited in their occurrence around the
world. Coking coals are primarily used to make coke that, under
high temperatures, reduces metal oxides to metals. This process
occurs when the coke is combined with the metal oxides at elevated
temperatures. The carbon from the coke combines with the oxygen
from the metal oxides to produce carbon dioxide, liquid metal and
210
COAL
residual ash (slag). The coals most suitable for producing coke
command the highest prices on the world market.
Sulphur content is always undesirable. Creating air pollution
when the coal is burned, sulphur emissions must be controlled with
expensive technologies. Laboratory analysis of sulphur content as
percentage of total weight of coal is typically adjusted for the heat
content of a ton of the coal for pricing purposes, as regulatory stan-
dards are based on how much sulphur is emitted per ton of coal
burned.
High-rank coals are high in carbon and therefore heat value, but
low in hydrogen and oxygen. Low-rank coals are low in carbon but
high in hydrogen and oxygen content.
Transportation
More than any other energy commodity, transportation costs are a
major component of the cost of fuel delivered to the end-user. This is
a simple result of coals high bulk and weight relative to its value.
The high cost of transportation and rigidities in the transport infra-
structure impact the markets for coal. Coals are typically priced
either free on board (FOB) at mine origin, or cost, insurance and
freight (CIF) at the consumers destination, with either the consumer
or producer responsible for arranging and paying for transportation
from or to that point. There are no intermediate collection points and
few wholesale marketing points. Train shipments are difficult, if not
impossible, to re-schedule and re-direct, so there is very little trading
of physical coal once it is en route to an ultimate destination, unlike
the vast amount of trading of oil tankers. Seaborne coal markets are
where the most active trading occurs, because of the greater flexi-
bility and relative low cost of moving a bulky item across the water
versus across land.
Coal mines are either surface (open pit) or underground.
Transportation from the mine can be done through a number of
modes, but again the low value-to-weight ratio makes minimising
the physical handling of coal the key to cost efficiency in transporta-
tion. Depending on distance and mode of transport, transport costs
for delivered coal range from 2070% of delivered price to the ulti-
mate consumer, a major component of the total cost of coal
procurement.
Coal can be moved directly from source to end-user via truck for
211
distances of less than 100 miles. For longer distances, rail or water-
borne transport is typically used. Coal can also be trans-shipped
from rail or truck into river barges or ocean-going vessels. For other
than international export, no more than two trans-shipments would
be used, as it is important that transportation mode changes add as
little cost as possible. Therefore, coal goes from mine to end-user
with few intermediate transactions.
Historically, coal sold under long-term supply contracts with less
trading than other commodities due to high capital costs mirrored
on both the production side (mine and transportation development)
and the use side (power plant construction). Since many of the
mines, transportation networks and generation plants have been put
in place and their capital costs are amortised, the economics allows
for shorter deals. In addition, consumers have learned to be much
more flexible in sourcing, which enables coals to compete among
each other and against other fuels. Consequently, markets have
become more dynamic. Trading and risk management tools have
also grown to match that flexibility. An increasing proportion of coal
is sold on the spot market and priced off of indexes. This is what has
stimulated the growth of derivatives trading.
Cheaply mined and having relatively low heat content (and also
low sulphur content), Powder River Basin coals are shipped by rail
from Wyoming to west coast ports and then on to Asia. Eastern US
coals can change modes several times, from mine by rail or truck to
river barges and then out to Europe through loading on ocean-going
vessels in the New Orleans area, or directly by rail to ports on the east
coast. Once sea-borne, coals from Australia and South Africa
compete with the US coals for markets in Europe and Asia. The
consumer purchases the coal based on a limited number of heat
content and quality variables against the price delivered to their
power plant. Thermal coal has become for the first time a truly world
commodity, a fact that is reflected in the growth of derivatives
trading.
Bituminous coal is typically much more expensive to mine, has up
to 50% greater heat content and thus significantly lower transporta-
tion costs, and can be environmentally friendly, commanding higher
price at the mine. As mentioned above, sub-bituminous coal, such as
from US Powder River Basin, has lower heat content and transporta-
tion costs as much as 50% greater with long, overland rail
212
COAL
transportation to end-users or export terminals but has low mining

costs due to thick seams of easily accessible coal through surface
mining, and often has lower sulphur content. This makes it
extremely competitive, even in world markets, and significant export
capacity on US west coast is under development.
MARKET STRUCTURE
Worldwide, most coals are priced on a per ton basis. In the US,
however, many utilities prefer to buy on a price based on heat
content rather than weight, in million Btu (MMBtu).
Prices are measured by many indexes that are transparent and
reliable, and have allowed the growth of derivatives trading based
on them. In the past, published prices rarely changed and were
totally unreliable for any contracting or trading. Little spot trading
occurred and long-term contracts included negotiations of many
factors, particularly free supply options for the buyer, which made
price comparison across time or contracts meaningless. For these
reasons, active physical and financial trading of coal was slow to
develop, but has become fully integrated into the energy risk
management environment.
A joint venture between an energy market news organisation,
Argus Coal Services, and a coal industry economic and management
consulting firm, IHS McCloskey, produces the API indexes, which
are the standard industry benchmarks. The main focus for activity in
the coal derivatives market is the API 2 index, which consists of an
average of the two firms price assessments for coal imported into
Amsterdam, Rotterdam and Antwerp, and includes CIF. Another
major index is API 4, which is the benchmark for coal exported from
Richards Bay in South Africa and also incorporates CIF. Argus esti-
mates that more than 90% of the worlds coal derivatives are priced
against these indexes. The list below describes the key indexes used
for international physical and derivatives coal business.
API 2 index: the industry standard reference price used to trade

coal imported into northwest Europe. The index is an average of
the Argus CIF Rotterdam assessment and McCloskeys north-
west European steam coal marker.
API 4 index: the price for all coal exported out of Richards Bay,
South Africa. The index is calculated as an average of the Argus
213
FOB Richards Bay assessment and McCloskeys FOB Richards

Bay marker.
API 5 index: the price for exports of 5,500 kcal/kg net as received
(NAR), high-ash coal from Australia. The index is calculated as
an average of the Argus FOB Newcastle 5,500 kcal/kg assess-
ment and the equivalent from IHS McCloskey.
API 6 index: this index represents 6,000 kcal/kg NAR coal
exported from Australia. It is calculated as an average of the
Argus FOB Newcastle 6,000 kcal/kg assessment and the equiva-
lent from IHS McCloskey.
API 8 index: the price for 5,500 kcal/kg NAR coal delivered to
south China. It is calculated as an average of the Argus 5,500
kcal/kg cost and freight (CFR) south China price assessment and
the IHS McCloskey/Xinhua Infolink south China marker.
The publishing schedule for these widely used indexes are as

follows:
Weekly average coal price:

Northwest Europe (CIF ARA) API 2 index;
South Africa (FOB Richards Bay) API 4 index;
Australia (FOB Newcastle) API 5 index;
Australia (FOB Newcastle) API 6 index; and
CFR south China API 8 index.
Monthly coal price: API 2, API 4, API 5, API 6, API 8 indexes; and
Daily coal price: API 2, API 4 indexes.
These prices are available exclusively through the Argus/

McCloskeys Coal Price Index service.
FINANCIAL MARKETS FOR COAL

Virtually all markets are served by multiple over-the-counter (OTC)-
cleared standardised derivatives contracts. Multiple platforms offer
products on the same indexes. OTC trades are cleared on two
competing platforms: CME/Nymex and the Intercontinental
Exchange (ICE). The types of coal futures for each exchange are listed
below (as of June 2013). The exchanges also list options and strips for
most of these futures.
214
COAL
CME coal product slate

Thermal coal products
Global:
MTF: Coal (API 2) CIF ARA (Argus/McCloskey);
s Bay (Argus/McCloskey);
215
Supply: mountaintop removal and water-course impacts for

mining; construction of transportation facilities such as major
rail improvements or development of export terminals.
Consumption: emission of carbon, sulphur oxides and nitrous
oxides on the consumption side; retrofitting of new control tech-
nologies and purchase of emission allowances and credits.
Supply and trade

Minerals mining companies focused primarily on coal extract the
majority of the produced coal. Such companies range from national
producers to international corporations, as well as many smaller
companies. While it used to be very common, particularly in the
Appalachian region of the US, for companies to be formed to own
and operate just a single mine, much of the industry has taken
advantage of economies of scale that have resulted in a greater
concentration of ownership in larger corporations. Consequently,
short-term spikes in price can occur due to strikes, labour shortages,
transportation bottlenecks and mining problems at the larger mines.
Typical production cost increases in major coal exporting coun-
Figure 8.1 Largest coal exporters annual exports (thousand short tons)
400,000
Indonesia
350,000 Australia
Russia
United States
300,000 Colombia
South Africa
250,000
200,000
150,000
100,000
50,000
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Source: EIA, international energy statistics
216
COAL
tries outside the US have increased by around 200% since the late
2000s. These dramatic rises in cost vary across production areas and
are due to a wide variety of reasons. The main impact has been to
increase the integration of worldwide coal markets as producers look
for more extensive markets and consumers search for competitive
purchasing opportunities.
Figure 8.1 shows the changing landscape of the top global coal
exporters. Almost half of Australias exported coal goes to metallur-
gical use, mainly in Asia and Europe, with Japan, India, China and
South Korea being the main Asian importers. Japan is also the largest
buyer of Australian thermal coal. The US and Canada export signifi-
cant quantities of metallurgical coal, but thermal coal comprises
most of Indonesias rapidly growing export volumes. China,
South Korea, India and Japan are the largest importers of US coal.
Figure 8.2 shows the distribution of recoverable reserves for coal
globally, while Figure 8.3 displays the trends for the largest
importing countries.
Consumption
While there are other important trends in coal demand, such as
continued growth in Indias consumption and imports, China alone
has dominated global consumption and demand growth. Again
Figure 8.2 World recoverable coal reserves (861 million tons)
Other US
26% 28%
Indonesia
6%
China Russian Federation

Australia 13% 18%
9%
Source: BP, June 2013, Statistical Review of World Energy
217
Figure 8.3 Largest coal importers annual imports (thousand short tons)
250,000
200,000
150,000
100,000
50,000
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Japan India
China Taiwan
South Korea
Source: EIA, international energy statistics
quoting from BP, Statistical Review of World Energy (June 2013)

regarding 2012 annual growth in coal use:
Consumption outside the OECD rose by a below-average 5.4%;
Chinese consumption growth was a below-average 6.1%, but China
still accounted for all of the net growth in global coal consumption,
and China accounted for more than half of global coal consumption
for the first time. OECD consumption declined by 4.2% with losses
in the US (11.9%) offsetting increases in Europe and Japan. Global
coal production grew by 2%, with growth in China (+3.5%) and
Indonesia (+9%) offsetting a decline in the US (7.5%). Coal reached
the highest share of global primary energy consumption (29.9%)
since 1970.
Thermal coal consumption in the US has decreased since 2008

compared with increasing consumption in Asia and Europe. In the
US, natural gas continues to displace more and more coal generation
due to the costs of upgrading old coal plants to meet ever-higher
emission standards, and the low cost of gas due to greater expansion
of supply.
The relationship between natural gas prices and coal prices for
power production has driven coal markets like never before. The
218
Figure 8.4 EIA historical and forecast annual coal consumption (quadrillion Btu)
100
90
80
China
70 United States
OECD Europe
60 India
OECD Asia
50 Rest of World
40
30
20
10
0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Source: EIA, international energy statistics database (as of November 2012), and EIA Annual Energy Outlook 2013 (base case).
COAL
219
glut of natural gas in the US in 2012 drove natural gas prices to a level
where, in July 2012, for the first time in history US electricity produc-
tion from gas-fired plants was equal to that of coal-fired plants.
Contrast this with the early 1990s, when coal represented more than
50% while gas represented roughly 5%. At prices above US$3.50/
MMBtu for gas, coal becomes competitive again. Gas-fired power
displaced US coal in the international markets, where the cheap coal
significantly increased European coal-fired generation at the expense
of their natural gas plants.
Chinas consumption growth comes largely from increasing
power generation. China has large domestic coal reserves, but it will
always take advantage of low import prices and significantly
increase imports appropriately. Since US demand has been down
due to the explosion of inexpensive supplies of natural gas, China
has imported US and other coals while reducing domestic produc-
tion. When demand and prices increase in the US domestic markets,
China will rely on its own production again.
PANEL 8.1: FUEL TO POWER SPREADS

A key ingredient in most liquid derivatives markets is the trading of spreads
between one instrument and another. In effect, most commodity trading is
based on the differential between two (if not more) prices. Few traders take
on outright risk, but most do choose very specific relationships where they
have developed expertise and expect that they can both recognise certain
trends before the market has fully taken them into account and can, in any
event, minimise the risk exposure made by each trade.
In options trading, there is a whole unique vocabulary describing the
various types of spreading between puts and calls on various strike prices
for the same security or commodity. Often, commodity futures spreads are
built on assessments of the likely trends in differentials between various
contract months in the same commodity. Classic commercial hedgers
spread the exposure between long and short positions for the same
commodity for the same delivery period, with one side being in the actual
purchase or sale of the physical commodity with an offsetting derivatives
position. In theory, whatever loss accrues on one side should be offset by
a corresponding gain on the other, keeping the producer, consumer or
merchandiser of the actual commodity protected against swings in the
market price of the commodity. Hedging thus frees up the firm to focus on
operational and marketing efficiencies rather than worrying about its busi-
ness being disrupted by volatile price movements. This is, of course, in
theory. In practice, the basis of the differential between the underlying
220
COAL
physical and the financial derivative must be managed closely. Hedging is

often called an exchange of absolute price risk exposure for basis risk
exposure.
Most pertinent to coal are the spreads between inputs and outputs in a
commodity production process. The classic spreads are the soybean
crush, whereby trades are made on the differentials between the raw
soybean inputs which are crushed to make soybean meal and soybean oil,
with each having their own derivatives. With the advent of petroleum
derivatives trading, the petroleum crack spread became a key differen-
tial, measuring the cost of raw petroleum being refined into heating oil and
gasoline. Since petroleum refiners often had catalytic cracking units, and
crack is similar to crush, the petroleum crack was the logical new
name.
With the advent of electricity, natural gas and coal trading, the deriva-
tives world added the spark spread, the differential between natural gas
fuel prices and electricity output prices. For coal-fired plants, the equiva-
lent spread is the dark spread. Both of the spark and dark spreads can be
called dirty when they do not incorporate the cost of purchasing carbon
credits for emissions created by the plants.
In all these input-to-output spreads, financial traders develop standard-
ised relationships describing the amount of each input required for each
unit of output. As the reality for each bean-crushing operation, each oil
refinery or electric power plant will vary from these standard trading
models, the hedging/risk management teams for those operators will
adjust their trades accordingly.
For a simple example of a dark spread calculation using US measure-
ment units, the spread is measured by:
Spread = [Power price (US$/MWh)] [Coal cost (US$/ton) +
Transport cost (US$/ton)] x [Heat rate of generator (MMBtu/MWh)
Heat content (MMBtu/ton)]
where MWh is megawatt hour, heat rate is the rate at which the electric
generator converts heat from the coal combustion into power, measuring
the efficiency of the generating unit, and heat content is how much heat is
produced by burning a ton of that coal.
Unlike spark spreads, which are calculated using natural gas costs and
on-peak power prices, dark spreads often use a combination of on- and
off-peak power prices. This combination (referred to as the flat price)
reflects the different role that coal-fired generators play in the supply stack
of a particular electric system. Coal-fired generators have traditionally
served as base-load generation. They run throughout the day and night.
The combination of on-peak (during the day) and off-peak (nights and
weekends) power prices reflects this role.
In addition, as gas-fired plants are typically more efficient than coal,
typical spark spread heat rates correspond to an efficiency of around 0.5
(50%), while dark spread heat rates are near efficiencies of 0.38 (38%).
221
CONCLUSION
Typical production cost increases in the major exporting countries
other than the US have increased by about 200% since the late 2000s,
resulting in the integration of worldwide coal markets.
Different coals compete with each other through a sometimes-
complex value optimisation, combining quality, suitability, location
and cost of transport. Quality differentials continue to play a bigger
role in import decisions for coking coal because they play a bigger role
in the suitability for various steel plants. This contrasts with steam
coal, which is basically just heat and is very interchangeable. There
are sufficient known and accessible reserves of met-quality coal;
however, due to the increases in production costs, prices have to rise
to bring them to market. Therefore, if demand for steel production is
sufficient, met coal prices will rise to meet the input demand.
Demand drivers are factors that move electricity demand such as
weather, economic growth and, to some extent, the price of
competing fuels including natural gas. Met coal demand depends
directly on steel production.
Multi-year coal contracts have been in a long process of evolution
since the early 1990s. It used to be fairly easy to describe typical terms
and conditions, but this is no longer the case as there are many types
differing within countries and from country to country.
Coal remains the single most important fuel for generating elec-
tricity worldwide. Traditionally, coal has been by far the cheapest fuel
for generating electricity. The other cheaper form is hydropower,
which is strictly limited by geography and annual weather conditions.
However, due to technological improvements in extracting natural
gas, that fuel has become consistently competitive to coal on price.
Furthermore, natural gas is less carbon-intensive than coal, its
burning produces fewer undesirable emissions and the capital costs
of building natural gas-fired generation are much less than for coal.
Therefore, coal has lost significant ground to natural gas. Due to its
abundance and the high level of installed generating capacity,
however, coal will continue to play a significant role in electricity
generation.
Since the beginning of the industrial revolution, coal has been and
continues to be the workhorse of the energy world. The coal
marketing chain from production to final consumer is typically much
less diverse and complex than other commodities. Coal also has a
222
COAL
much lower value per weight than other commodities. Also, indus-
trial organisations are the exclusive end-user consumers for coal. The
high proportion of transportation costs and less-diverse end-users
result in few transactions from mine-mouth to final consumer.
Therefore, among the major energy commodities, coal markets have
been the slowest to adopt financial derivatives. However, coal has
become a full member of the energy risk management jigsaw.
APPENDIX 8.13
Coal conversion statistics and terminology
Basis of analysis
Definitions:
as received (ar): includes total moisture (TM);

erent moisture (IM) only;
To obtain: Air dry Dry basis As received

multiply
ar by: (100 IM%)/(100 TM%) 100/(100 TM%)
ad by: 100/(100 IM%) (100 TM%)/(100 IM%)
db by: (100 IM%)/100 (100 TM%)/100
[For daf, multiplydb by 100/(100A)]
Example:
ar ad db daf
TM 11.0
IM 2.0 2.0
Ash 12.0 13.2 13.5
VM 30.0 33.0 33.7 39.0
FC 47.0 51.8 52.8 61.0
Sulphur 1.0 1.1 1.12
223
MASS
Units:
Metric ton (t) = tonne = 1,000 kilograms (= 2,204.6 lb);
Imperial or long ton (lt) = 1,016.05 kilograms (= 2,240 lb); and
Short (US) ton (st) = 907.19 kilograms (= 2,000 lb).
Conversions:
From long ton to metric ton, multiply by 1.016;
From short ton to metric ton, multiply by 0.9072;
Mt million tonnes;
Mtce million tonnes of coal equivalent (= 0.697 Mtoe); and
Mtoe million tonnes of oil equivalent.
Calorific values (CV)

Units:
kcal/kg Kilocalories per kilogram;
MJ/kg* Megajoules per kilogram; and
Btu/lb British thermal units per pound.
* MJ/kg = 1 Gigajoule/tonne (GJ/t)
Gross and net calorific values

Gross CV or higher heating value (HHV) is the CV under labora-
tory conditions.
Net CV or lower heating value (LHV) is the useful calorific value
in boiler plant. The difference is essentially the latent heat of the
water vapour produced.
Conversions (units):
From kcal/kg to MJ/kg, multiply by 0.004187;
From kcal/kg to Btu/lb, multiply by 1.800;
From MJ/kg to kcal/kg, multiply MJ/kg by 238.8;
From MJ/kg to Btu/lb, multiply MJ/kg by 429.9;
From Btu/lb to kcal/kg, multiply Btu/lb by 0.5556; and
From Btu/lb to MJ/kg, multiply Btu/lb by 0.002326.
Conversions gross/net (per ISO, for as received figures):

kcal/kg: Net CV = Gross CV 50.6H 5.85M 0.1910;
MJ/kg: Net CV = Gross CV 0.212H 0.0245M 0.00080; and
Btu/lb: Net CV = Gross CV 91.2H 10.5M 0.340.
224
COAL
where M is percentage moisture, H is percentage hydrogen, O is

percentage oxygen (from ultimate analysis,4 also as received).
For typical bituminous coal with 10% M and 25% volatile matter, the
differences between gross and net calorific values are approximately
as follows:
260 kcal/kg 1.09 MJ/kg 470 Btu/lb
Power generation:
1 MWh = 3600 MJ;
1 MW = 1 MJ/s;
1 MW (thermal power) [MWth] = approx 1,000 kg steam/hour;
th/3.
1 Aaron Blake, Washington Post, June 25, 2013: Obama science adviser calls for war on coal.
2 BP, 2013, Statistical Review of World Energy, June.
3 Source: World Coal Association website: http://www.worldcoal.org/resources/coal-
statistics/coal-conversion-statistics/.
4 Ultimate analysis determines the amount of carbon, hydrogen, oxygen, nitrogen and
sulphur.
225
Part II
Trading and Investment

Strategies
9
Farmland as an Investment
Greyson S. Colvin and T. Marc Schober
Colvin & Co. LLP
Although oil, metals, grains and financials are commodities key to

making the world go round, only farmland has no substitute.
Everyone has to eat in order to survive, and the production of almost
all food can be traced back to farmland. Demand is growing for farm-
land as the worlds population and global need for food increases.
However, what many do not realise is that the supply of farmland is
not changing, thus creating a severe imbalance in its supply and
demand.
Over the long term, farmland will provide a steady stream of
income and capital gains due to the increasing global demand for
agricultural commodities, driven by the rising world population,
rapid growth in emerging markets and continued demand for
ethanol and bio-fuels.
To understand it properly, we have to ask what exactly is farm-
land? The definition of farmland or agricultural land is the land
suitable for agricultural production, both crops and livestock.
According to the United Nations Food and Agriculture Organization
(FAO), there are three primary types:
1. arable land: land under annual crops, such as cereals, cotton,

other technical crops, potatoes, vegetables and melons; also
includes land left temporarily fallow;
2. orchards and vineyards: land under permanent crops (eg, fruit
plantations); and
3. meadows and pastures: areas for natural grasses and grazing
of livestock.
229
For our purposes, we will generally focus on arable land or row crop
farmland that produces grains planted in rows harvested each year,
including corn, soybeans and wheat. These are the grains that are
(and will be) needed to feed the worlds growing population. We
will also look at farmland located in the US, since it has some of the
best producing farmland in the world, as well as the most advanced
farmers and farming technology, the most developed infrastructure
and uses the most leading technologies.
According to the Natural Resource Conservation Service (NRCS),
there are 12 recognised types of soil in the world. Of these, the most
naturally fertile are mollisols, which is suitable or very suitable farm-
land. Mollisols are generally found in only four places: in the Pampas
Region of Argentina, the Steppes of Ukraine and Russia, areas of
Northeast China and the Grain Belt of America. Mollisols make up
only 7% of the ice-free land in the world and are the best soils for
farming because they contain large quantities of organic matter.
Mollisols found in the Midwestern US are the best for agriculture
due to the grasslands formed thousands of years ago. These prairies
produced strong and fertile soils because each year the grasses (and
animals) would break down, with nutrients in the organic matter
decomposing into the ground. Once the Wisconsin Glacier retracted
from Illinois and Iowa, great dust storms blew fertile silt on top of the
young land, making it ideal for crop production.
However, in terms of percentage of land area, not very much of
the planet is actually appropriate for farming. Once you remove
places that are too cold or too wet, the deserts, the forests, the bad
soils and every other strange place that cannot host a decent haul of
crops, there is not much left over. However, while America has 5% of
the worlds population, half of its land is suitable for cultivating and
growing crops. In comparison, China has 20% of the worlds popula-
tion but only 7% farmable land, according to the FAO.
Under the rule of law, US farmland cannot be hijacked by a totali-
tarian government or organised crime (yes, organised gangsters do
terrorise and control some farms in the Ukraine and Russia), and the
US Midwest Corn Belt sits in the optimum climate for production.
When coupled with modern technology, the US farmers work ethic,
excellent soil and infrastructure for transporting crops, the US is
unsurpassed for production.
All farmland is not created equal and no two properties are the
230
FARMLAND AS AN INVESTMENT
same. The ability of the land to produce profitable crops is part art
and part science; however, at the end of the day, so is analysing and
valuing farmland. This chapter will therefore cover the following
factors that drive the fundamental investment rationale for farmland
investments.
Land scarcity: there are approximately 3.5 billion acres of arable

ng a mere 5% over
for proteins will
231
Cash returns: farmland is a performing asset, generating modest

cash returns of 46%, depending on location and crop.
Sustainable asset: farmland improves in productivity over time
when well managed.
The chapter is organised into the following sections: the first will
examine value creation and investment in farmland, before we delve
into renewables and their impact. The next section details production
and its limitations, and we finish with an investigation into global
farming.
VALUE CREATION AND INVESTMENT

Value creation from farmland
Arable land for farming has been valued since the first crops were
domesticated. Farmland creates multiple commodities: wheat, corn,
animal products and meats, and even wind energy if a landowner
chooses to lease out part of their land to host a wind turbine. An
investment in farmland can provide a steady stream of income from
demand for agricultural goods, driven by the rising world population
and rapid growth in emerging market consumption. The continued
demand for ethanol and bio-fuels also puts upward pressure on crop
values. Demand for agricultural commodities is outpacing supply,
which positions farmland for long-term appreciation.
We should look at what makes something valuable as a
commodity; is it, or does it offer, a broadly desired marketable item?
Is it something that would be dearly missed if it disappeared from
the worldwide market? In addition to being an end-user item, can
something also serve as an investment vehicle?
Farmland as lease property

A farmland owner who does not intend to operate the farm often
monetises the lands value by leasing. A rental lease, in this case, is an
agreement between the landlord and the farmer of the property.
Often these agreements are legally binding documents drafted by
attorneys, but can be as vague as a verbal commitment (in which
case, it may be as solid as the paper it is written on).
Leases span all different lengths of time, from one year to the life
of the property, but in the Midwestern US they are often for between
one and five years. Farmers aggressively seek leased land for their
232
operations in order to expand and capture economies of scale

without increasing the most expensive element of production: the
land. Leased land allows farmers to spread equipment and other
fixed operation costs over more acres to increase profit margins, and
also allows them to increase income by farming more acres.
There are several possible lease options available to a landlord, but
any of them should return roughly a third of all revenues generated
from the land per year. There are three main types of farmland
leases:
cash rent: fixed rate per acre per year;

wner shares in the expenses and profits; and
Figure 9.1 Farmland riskreturn profile
Low risk High risk

Cropland Prime farmland Timber Tree crops
Low returns High returns
Own/hold Cash rent Crop share Custom farm Joint venture Operate
3% 5% 7% 10% 12% 12%+
233
landowner from taking on crop or credit risk from the farmer. The
landowner does not have to worry about drought or the rate of crop
growth. Land across the Midwest is typically leased at 45% of the
market value of the land; target farmland for investment that can be
leased for 5% or greater is recommended. Farmland in other regions
of the US can have lower lease rates as a percentage of value due to
the commodities produced and other factors affecting the value,
such as potential development.
Farmland as an investment
Farmland has a proven record it has been one of the top performing
investments over the last 100 years. In the 20th century, farmland
only decreased in value three times: during the Great Depression, the
inflation crisis of the early 1980s and in the housing crisis of 2008/09.
The US farm sector has a healthy balance sheet and, as mentioned,
debt-to-asset ratios are low. Unsurprisingly, farmers historically
have been the main buyers of US farmland and do not buy intending
to flip for profit but rather to hold for decades or generations,
keeping the land in the family. Farmland is the most valuable asset a
farmer can own, which leads most to reinvest a significant part of
their crop and livestock revenue back into the purchase of additional
farmland to expand their operations.
It is also important to understand that farmland values per acre
are essentially a function of revenues generated per acre. Revenues
are mainly dictated by two variables: price of the commodity and
yield per acre. In the 20th century, grain prices were fairly stable
while production increased a few percentage points per year, on
average. The increase in production allowed farmland to become one
of the most stable and consistent asset classes.
Despite three downturns over the last 100 years, farmland returns
in the US are historically one of the best investment vehicles,
comparing favourably with more traditional assets such as stocks
and bonds. Table 9.1 clearly shows the stability of farmland. Bear in
mind, this includes crop years and/or regions that were wiped out
or suffered severely diminished yields due to drought, flood and
other disasters.
In 2012, the Federal Reserve Bank of Chicago reported that farm-
land values grew by 16%, the third largest increase in the previous 35
years. Despite the worst drought in over 55 years, high commodity
234
Table 9.1 Midwestern US farmland returns
State 1 year 5 years 10 years 20 years 50 years 100 years

(%) (%) (%) (%) (%) (%)
Illinois 22.8 12.0 11.8 8.1 6.9 4.6

Iowa 22.8 16.2 14.1 9.7 7.6 4.9
Nebraska 33.5 18.4 13.5 8.7 7.4 4.6
North Dakota 26.5 14.1 11.8 7.3 6.8 4.2
South Dakota 23.9 13.1 12.8 8.5 7.1 4.1
Wisconsin 7.4 3.7 7.4 8.5 7.4 4.7
US 10.9 5.8 8.3 6.9 6.5 4.5
Source: USDA Economic Research Service
prices and record farm incomes drove demand for agricultural land.
Survey respondents anticipated that the momentum would continue
over the next 12 months based on the record income expectations for
2013. Iowa farmland values led the pack, with a 20% return in 2012,
followed by Illinois and Michigan with an 18% annual return. This
was during a time many considered recessionary.
One of the most attractive attributes of farmland is income
realised from rental. Since 1967, rural cash rents have yielded
roughly 5.7%, according to the USDA (this was calculated by the
authors using historical data from: http://usda.mannlib.cornell.
edu/MannUsda/viewDocumentInfo.do?documentID=1446). This
compares very favourably to Treasury bonds and other income-
producing assets. The cash rental contract is typically prepaid, so the
investor does not have to take operational or credit risk from the
farmer. Society will undoubtedly be drastically different by the mid-
21st century, but the US farmer will still be leasing farmland to raise
livestock and crops.
Farmland also provides investors with the chance to diversify
from traditional investments, which makes it an excellent asset to
balance a portfolio and offset financial and commercial real estate
market volatility. Farmland has always shown a positive correlation
to the Consumer Price Index (CPI), exceeding stocks, bonds and non-
farm real estate.
Farmland is frequently compared to investing in gold because of
its characteristic as an inflation hedge. However, unlike gold, farm-
land also produces a stable income stream, and as a consequence it
has been described as gold with yield. Gold does not stock-split or
235
Figure 9.2 Investment correlation with farmland (19712009)

Historical correlations with US Farmland
Correlations Negative Positive
Long term US corporate bonds -0.43
US treasury bills -0.22
S&P 500 -0.18
International equities -0.15
US small cap equities -0.07
US commercial real estate +0.23
S&P GSCI +0.28
Gold +0.30
US inflation +0.36
-0.50 -0.25 0 +0.25 +0.50
Source: NCREIF, Ibbotson & Associates, Morningstar, Western Spectator (June 2010)
pay dividends; you just hang on to it, pass it down or sell it. It can
also be seen as similar to non-dividend paying equities. Eventually,
the only way these stocks bring value to you or your family is when
you sell them. However, farmland will bring returns to you and
generations of your family as long as they continue to own and
manage the land.
RENEWABLES AND THEIR IMPACT

Renewable fuels impact on the farm
Social and political concerns regarding climate change and fossil-fuel
dependency have led to a significant focus on renewable fuels, such
as ethanol, as a replacement for petroleum-based fuel sources.
Ethanol is primarily manufactured from crops such as corn, wheat
and sugar cane. According to the USDA, ethanol production in the
US increased from less than three billion gallons in 2003 to over six
billion gallons in 2007, and is estimated to exceed 12 billion gallons
by 2020. The Renewable Fuel Standard from the 2007 Energy
Independence and Security Act calls for total renewable fuel to reach
36 billion gallons by 2022.
Ethanol, no matter how viable or controversial, is mandated as a
renewable source of energy. At its most basic, ethanol is grain
alcohol, produced primarily from corn and sugar cane. The USDA
estimates that more than 40% of US corn production was used to
produce ethanol in 2011. In January 2011, the US Environmental
Protection Agency (EPA) approved the use of E15 gasoline for vehi-
236
cles manufactured in 2001 or thereafter. Almost all gasoline in the US

is E10, or 10% ethanol content. The increase to E15 will help the US in
its goal of using 36 billion gallons of renewable fuel by 2022, as per
the Energy Act of 2007.
In 2004, the US government passed a 45 cents-per-gallon tax
credit, commonly known as the blenders tax credit, to provide an
economic incentive to blend ethanol with gasoline. The official name
is the Volumetric Ethanol Excise Tax Credit, and it was part of the
American Jobs Creation Act of 2004, although the incentive expired
at the end of 2011. In response, critics have argued that ethanol is an
inefficient source of energy and should no longer be supported by
the government. However, it seems unlikely that ethanol production
will disappear in the near future. The federal government does not
look to be changing these mandates.
Wind energy is another source of commodity revenue for the rural
landowner. By its very nature, farmland usually lies in the vast
expanses of open prairie that allows the winds unfettered flow.
Wind energy could even meet 20% of the US electricity demand by
2030. According to the US Department of Energy (DoE), farmland
owners can benefit from wind energy by having one or more wind
turbines placed on their property and receive a lease-rate payment
per turbine.
Landowners can receive up to US$15,000 annually per turbine,
although each wind companys contract will differ. One wind
turbine only requires roughly a single acre of land and has minimal
effect on farming practices. One acre of cropland is lost, but is
replaced with revenue from wind turbine leases. Once the wind
turbines are finalised and constructed, landowners typically receive
fixed and variable payments based on electricity production. South
Dakota is in an excellent position to capitalise on wind energy, as the
state is known as the Saudi Arabia of Wind.
According to Dakota Wind Energy, South Dakota has the wind
potential to meet 50% of US electricity demand. It ranks fourth in the
nation in wind power, behind North Dakota, Texas and Kansas.
Since the late 1980s, the cost to produce wind electricity has dropped
a huge 90%, according to the American Wind Energy Association
(AWEA). Although wind energy costs are not as low as for conven-
tional power, ever-improving technology is driving wind energy
costs down. The government has helped promote the development
237
of wind energy through subsidies, such as accelerated depreciation

and the production tax credit (PTC), which offsets the cost of devel-
opment.
The primary constraint of wind energy is the transportation of
electricity. Since electricity must be used immediately or transported
to a power plant, wind turbines must be closely connected to electric
grids that can transmit the energy. The majority of the windy regions
of the US are located in rural areas with limited amounts of energy
demand and transmission capacity.
One solution is the Green Power Express transmission line being
developed by ITC Holdings Corporation. The transmission lines
would span roughly 3,000 miles from the Dakotas into Wisconsin,
Illinois and Indiana. The Green Power Express, due to be completed
by 2020, will provide a path for newly generated electricity to travel
to heavily populated areas such as Milwaukee and Chicago, and
even open up the entire eastern seaboard. This may very well
involve an opportunity for landowners to lease land for infrastruc-
ture development in support of the initiative.
Fuels based on crops may be new, but a windmill on a US farm is
as old as a Norman Rockwell painting. Farms started featuring
windmills on their properties as early as 1900 for the purpose of
powering the well pump. It was not electricity, but the mill gener-
ated power and reduced the need of human or animal power
through harnessing natural wind energy. Efficiently introducing the
new technologies of wind turbines and eco-fuels allows a landowner
to even further diversify the sources of revenue from their farming
enterprise.
Increases in demand for agricultural products

Grain supplies in the US and globally are at decade lows, driven by
emerging market demand, disappointing US yields and demand for
bio-fuels. The ending corn stocks-to-usage ratio has been trending
downwards, from roughly 20% in 2004 to 5.6% in 2012, according to
the USDA (these figures were calculated by the authors; the data on
which this is based can be found here: http://www.usda.gov/oce/
commodity/wasde/).
In the USDAs October, 2012, update of World Agricultural
Supply and Demand Estimates report, ending stocks for 2012/13
are projected to be down by 37% to 619 million bushels, as corn use is
238
Figure 9.3 US corn stocks/usage ratio

70%
60%
50%
40%
30%
20%
10%
0%
1980 1986 1992 1998 2004 2010
Source: ERS/USDA
expected to exceed production by 444 million bushels and the

Midwestern US has had the worst drought in over 50 years. US corn
stocks have declined to a 21-day supply, meaning that if corn
production was halted, the US would run out of corn in a little over
half a month.
The global demands for food and rising commodities prices have
driven agriculture fundamentals upwards. The USDA estimates that
farm incomes have been steadily trending higher, increasing from
28% in 2010, 47% in 2011 and was recorded at 14% in July 2013, and
will continue to rise allowing farmers to reinvest their dividends
back into farmland to expand their operations.
Despite the rapid growth in agriculture, farmers balance sheets
remain very conservative. Strong farm income and minimal use of
Figure 9.4 Farm sector debt-to-assets ratio

25
20
15
10
0
1960 1970 1980 1990 2000 2010
Source: ERS/USDA
239
debt have allowed the US farm sector to maintain conservative

balance sheets as current debt-to-assets ratios continue at decades-
long lows. New banking regulations have restricted the access to
capital for farmland buyers, and loans secured by farmland are typi-
cally limited to 50% of the purchase/appraised price. This secure
financial situation bodes well for farmland (farmland owners tend to
be more on the commonsense side of economics).
PRODUCTION AND ITS LIMITATIONS

Limits to production
Farmland values are expected to continue their momentum into the
2020s and beyond due to the strong global and ever-increasing
demand for food. The worlds gross agricultural output must
increase by 3.4% to meet this demand, according to the FAO. The
two primary ways to increase agricultural production are to either
increase the amount of acres planted or increase productivity with
technology. With urban sprawl and land development, increasing
yield seems to be the logical answer.
The future ability to expand arable acres will be difficult. The
prime areas for farming have already been identified, are being used
for production and have built-in transportation and infrastructure
support. The marginal arable acres that can be put into production
will be in odd, out-of-the-way places with less than optimal growing
conditions and possible transportation issues. However, there is a
way to grow yield and increase arable acres.
Although the introduction of genetically modified organisms
(GMOs) has been somewhat controversial, they have not only
increased bushels per acre in standard farming regions, but they
have also brought better drought and cold tolerance in the US, as
well as expanding the land area that can be used for cold-sensitive
crops. For instance, the land planted to both corn and soybeans since
the late 1990s has extended into the colder north and drier west
areas. The acreage allotted to corn and soybean production is
expanding northwest to regions where the number of growing
degree days are less. Crop insurance for corn acreage now expands
60 miles further north into Canada. As a result, the Corn Belt and,
along with it the opportunity to invest in high-quality producing
farmland, continues to grow.
240
Acres in conservation programmes

There is yet one more resource for production: farmland set aside
under the Conservation Reserve Program (CRP) could be added to the
amount of US arable acreage. According to the USDA, 31.3 million
acres had been enrolled in the CRP under almost 738,000 contracts by
the end of 2010. As the CRP contracts expire, much of this land may be
put back into production, but a majority is marginable at best, which is
the primary reason it was put into the programme in the first place.
The CRP pays landowners not to farm their cropland in order to
protect areas where wildlife can grow and fertile land can take a break
from producing crops. Other environmental programmes include
environmental quality incentives and wetland preservation. This
must be done for the long-term health of the soil. CRP will provide
more acres in the US for production, but due to the lack of soil quality,
the effect on total production will be minimal.
GLOBAL FARMING
Farm growing in other global regions
The amount of acres of arable farmland has been almost static as the
non-farm development of farmland in North America and Europe
has been offset by expansion of farmland in Africa and South
America. There are approximately 1.5 billion hectares being farmed
around the world. The FAO estimates that the world has a total of 2.5
billion hectares of very suitable or suitable land for farming and
raising crops. About 80% of this reserve land is located in Africa and
South America. The investment bank Credit Suisse estimates that
there is only about 300,000 hectares of additional potential acreage,
with the majority in Brazil and Indonesia.
Table 9.2 summarises the acres in use in 2013 and potential global
arable acres. The primary expansion opportunity lies in Brazil,
where the government organisation Conab estimates there are an
additional 106 million hectares available for agricultural develop-
ment. Historically, the soil was thought of as unfarmable due to high
acidity levels and lack of nutrients. However, technologies such as
strip tilling, soil surveys and Global Positioning Systems (GPS) have
allowed farmers to improve soil fertility, and a new type of soybean
developed to grow in tropical climates from the early 1980s meant
that farmers were able to start producing crops in previously unsuit-
able acreage.
241
Table 9.2 Global acreage expansion
Area (1,000 Ha) Acres (2013) Additional acres
Europe 94,294 1,000

US 173,158 12,950
Brazil 66,500 106,000
Other Latin America countries 59,290 76,000
Indonesia 37,500 102,000
Russia 123,368 10,397
Ukraine 33,333 1,120
World total 1,553,689 309,467
% of world total 20%
Source: Conab, Indonesia Ministry of Ag, USDA, FAO, Credit Suisse
Indonesia has a huge opportunity to expand acreage for palm oil

cultivation. The Indonesian government estimates that it is only
using half of its land available for cultivation. In January 2011,
Indonesia targeted expanding the countys agricultural land by two
million hectares in the medium and long term, although this plan has
received a great deal of criticism as it would result in the removal of
tropical forests.
Ukraine, Russia and Kazakhstan saw a substantial decline in
arable acres and crop yields following the decline in communism
during the early 1990s. This demonstrates the loss of the motivation-
to-yield prospect of farming: farming is hard work and if your labour
goes into the pockets of organised crime or corrupt government,
there is no incentive towards healthy crop production. The FAO esti-
mates that arable acres declined 11% between 1992 and 2005. Credit
Suisse estimates that if arable acres return to 1992 levels, that it
would add 1.9% to the total global arable acres.
Big (farm) trouble in China

There has been much speculation, and even fear, about the rise of
China. Chinese demand for agricultural products will likely be a key
force in these markets for the coming decades. The year 2010 marked
a new era for China as it announced it would no longer be self-
sufficient in corn production. The demand of the most populated
country in the world for corn and feed is now outpacing supply as
the nation continues to consume more and more protein. China and
its people are in the process of transitioning from a grain-based diet
242
to a protein-based diet. On average, it takes seven pounds of grain to

produce just one pound of meat, according to the Earth Policy
Institute.
One of the primary problems limiting Chinas ability to feed itself
is its land imbalance. China has roughly 20% of the worlds popula-
tion with only 7% of the worlds arable land. The supply of arable
farmland in China is decreasing rapidly as well. By 1950, China had
lost a fifth of its arable land due to erosion, desertification and devel-
opment, and is expected to lose 1015 million more hectares by 2020,
according to the UN.
In order to be self-sufficient in grain production, the vice minister
of agriculture, Wei Chaoan, stated in 2010 that China needed to
maintain 120 million hectares for crop production until 2020.
Government figures estimate that the amount of arable land is actu-
ally 122 million hectares, which has remained unchanged since 2005.
Bank of America estimates that Chinas arable land has already fallen
below the 120 million hectare threshold and could decrease to 117
million hectares by 2015.
As its economy and population grow, China will have to increas-
ingly rely on the import market to solve their shortage of corn and
other foodstocks. Chinese imports of corn will grow from 1.0 million
tons in 2010 to 15 million tons in 201415, according to the US Grains
Council. 15 million tons of corn translates to Chinese imports of 600
Figure 9.5 China corn supply demand
180,000
Production (1000MT)
140,000
100,000
60,000
20,000
80/81 90/91 00/01 10/11
Total production Total consumption Ending stocks
Source: USDA Foreign Agricultural Service
243
million bushels, equal to 15 tons of corn, will have a substantial

impact on global corn stocks.
Chinas transition to a net importer of corn is very similar to its
transition to becoming a net importer of soybeans. Before 1995,
China was a net exporter of soybeans, but by 2010 it was the worlds
largest soybean importer, importing more than 57 million tons of the
crop, 21.6% of world production, according to the USDA. The rapid
industrialisation of developing markets will have serious repercus-
sions on the demand for grain. In China specifically, there may be
around 500 million more people demanding a protein-based diet.
China is not the only example of a developing country that has
an increasing appetite for grains. As the worlds middle class
continues to develop, the demand for grains will continue to grow
exponentially.
Global demand for farm crops and commodities

According to the US Census Bureau, there were approximately 7.0
billion people inhabiting the Earth in 2012, compared to just 1.7
billion in 1900 and 5.8 billion in 1985. The rate of population growth
is not expected to temper as the United Nations estimates the worlds
population is likely to reach 9.2 billion by 2050. Most of this popula-
tion growth is expected to originate in emerging economies, with
developed countries remaining stable.
The global population growth rate is expected to decelerate due to
lower fertility rates, to roughly 1% by 2030, down from a 2% annual
growth rate in 1970, according to the United Nations. Despite the
slower population growth rate, life expectancies have substantially
improved from 3040 years in pre-industrial times, to roughly 65
years. The prospect of feeding a demographic that is becoming less
productive is another factor that puts a strain on food production.
In order to feed the worlds growing population, agricultural
output will need to double by 2050, according to the FAO. This will
be a daunting goal to accomplish as agricultural resources are
already strained. Since the early 2000s, agricultural output has
grown by 2.4% annually. In order to double agricultural output by
2050, output must increase at 3.4% per year. To meet future demand,
experts are predicting that global agriculture will need to produce
more food in the next 50 years than what was produced during the
previous 10,000 years, putting more and more pressure on future
farmers and the land they use to produce our food.
244
Figure 9.6 World population (19502050)

Billions
10
0
1950 1970 1990 2010 2030 2050
Source: US Census Bureau, International Data Base
Food demand is growing faster than population growth because

of the development of middle classes in emerging markets, due to
above-average GDP growth. The Brookings Institution estimates
that, by 2021, Chinas middle class could grow to over 670 million,
compared to only 150 million in 2010. Economists have long shown
that, as GDP rises, so does the consumption of animal protein as a
percentage of diet. As emerging economies continue to develop,
there will be a transfer from a grain-based diet to a protein-based
diet. Over half the increase in global calorie consumption since the
early 2000s has been a result of increased meat consumption,
according to the FAO. It takes two pounds of grain to produce one
pound of chicken, five pounds of grain to produce one pound of
pork and seven pounds of grain to produce one pound of beef.
Again, this represents a great demand for commodity production.
SUMMARY
Farmland values have been steadily increasing due to increased
commodity production on farmland, but the primary driver of
future value increases will derive from the supply and demand of
the commodities grown from the land. Corn supplies are at their
lowest levels in decades. The major difference between the 1995 corn
supply and corn supply in 2013 is that global corn production was
low in the mid-1990s due to poor production, which was only a
short-term effect. That US corn supply has become an alarming 20
days is due to the increased usage of corn across the entire world.
What is exciting about farmland is that the agriculture proposition
245
is still the tip of the iceberg. Most agriculture investors are attracted
to the sector because of the wealth creation due to the transfer to a
protein-based diet in emerging markets. China is expected to
increase corn imports from 1 million tons in 2010 to 15 million tons
by 2014. The biggest demand for grain by the emerging markets has
not even occurred yet. The basic supply and demand is in place for
farmland to continue its bullish trends in the long term.
Although the amount of farmland is limited in the US, farmable
corn-producing land is expanding into areas with great soil but
heretofore slightly unsuitable climates in the Midwest, primarily due
to biotech seeds. Large seed and agrichemical companies have
focused years of research on higher performing varieties and hybrids
of important food and feed crops. The next generation of biotech
traits focus on greater productivity, improved nutrient use, disease
resistance, plant density and drought and cold tolerance.
While GMOs may bring a degree of controversy, they also
generate much-needed crop acreage and yield. And with people
always looking for safe places to invest, this can translate to a great
investment upside through increased commodity production.
Although farmers make up the majority, people from many
different walks of life own farmland, and outside investors have
always had a minority interest. However, outside investor interest
has grown latterly and will keep growing as farmland continues to
feed the worlds growing population. Almost 200 investment firms
are expected to invest US$30 billion in farmland by 2015, according
to Michael Kugelman of the Woodrow Wilson International Center
for Scholars. Worldwide media coverage now includes farmland on
a daily basis and the expansion of farmland as an asset class
continues to occur.
The average age of the US farmer is steadily increasing. The 2007
Census of Agriculture reported their average age had increased from
50.3 in 1978 to 57.1 in 2007. The ageing farmer may provide an oppor-
tunity for the non-farmer investor to get into this commodity-
producing market. There was a time when the family farm went to
the son when the father retired or passed on. However, societal
trends have seen people selling the family farm and getting out of the
family business.
Demand is growing for farmland as the worlds population and
global needs for food increase. What many do not realise is that the
246
supply of farmland is not changing, thus creating a severe imbalance

in its supply and demand. An investment in farmland over the long
term will provide a steady stream of income and capital gains due to
the increasing global demand for agricultural commodities, driven
by the rising world population, rapid growth in emerging markets
and continued demand for ethanol and bio-fuels. Demand for agri-
cultural commodities is outpacing supply, which positions farmland
for long-term appreciation.
247
10
Agriculture Trading
Patrick OHern
Sugar Creek Investment Management
The pool of participants trading in agriculture commodities has

grown rapidly in number and type since the beginning of the 21st
century, thus increasing diversification and liquidity across the agri-
culture sector. Increased participation has been witnessed across
each subset of traders, including the commercial, non-commercial
and index-trading communities. This growing diversification across
agriculture markets has raised the bar for money managers and
proprietary traders alike who are seeking to exploit positive risk
reward opportunities. This chapter will provide descriptions of these
types of traders, their behaviour and objectives. This chapter is
arranged into three sections, which look at, respectively, the partici-
pants in the agriculture markets, trading in these markets and the
strategies utilised.
PARTICIPANTS IN AGRICULTURE MARKETS

Commercial traders
It is important to consider the commercial subset of traders, and
better understand their activities and objectives. Commercial traders
as defined by the US Commodity Futures Trading Commission
(CFTC) are those who use futures or option contracts in a given
commodity for hedging purposes. Commercial traders hold posi-
tions in both the underlying commodity and in the futures (or
options) contracts on that commodity. In agriculture, commercials
can be producers, merchants and end-users, all of which come to the
market to manage business, price and margin-related risks.
249
Commercial trading activity has grown to become more sophisti-

cated over time, as businesses have dedicated more capital to build
out trading desks by instituting structured commodity marketing
and risk-mitigating hedging plans for themselves and their
customers.
Figure 10.1 illustrates the growth in commercial participant
volumes traded across agriculture markets since the year 2000. The
expansion among the commercial trading community is viewed as
imperative as the globalisation of agriculture commodities has
increased the volatility in profit margins for all types of physical
commodity businesses. The increased volatility in profit margins has
driven commercials to put more emphasis on managing margin risk.
For instance, consider a large livestock feeding operation that takes
part in purchasing, feeding and selling the stock. The focus for this
operation is not only on hedging or marketing the sale price, but also
the purchase price and the input costs, including feed and energy
usage. Profit margins can vary greatly over the ownership period
due to changes in the price of input costs that can create enormous
business risks for the producer. For non-commercial traders, it has
become increasingly important to understand the behaviour and
underlying economics of these commercial trading entities, as the
business risk imbedded within participants such as the livestock
feeder are just as crucial as the supply and demand of the commodity
itself.
Figure 10.1 also illustrates the difference in the level of participa-
tion between commercial and non-commercial participants. This
difference highlights the importance for non-commercial partici-
pants to be more aware of the business and economic decisions being
made by commercial market participants, as they generally account
for 5060% of the aggregate trading volume and total open interest
across agriculture markets. In commodities, open interest is the total
number of futures and/or options contracts in a contract month,
while total open interest accounts for the total amount of contracts
across the forward curve per commodity.
Generally speaking, the non-commercial participation ranges
around 4060% of the commercial participation. As seen in Figure
10.2, CME Feeder Cattle non-commercial volumes are larger than
that of commercial volumes. This is due to an unusual amount of
commercial hedging activity falling into the non-reportable category,
250
Figure 10.1 Growth in commercial and non-commercial trading across agriculture markets
CBT wheat, KCBT wheat, corn, MGE wheat, oats, soybeans, soybean oil, soybean meal, cotton, rough rice, orange juice, milk, lean hogs,
live cattle, feeder cattle cocoa, sugar and arabic
7,000,000
6,000,000
5,000,000
Contracts
4,000,000
3,000,000
2,000,000
1,000,000
0
1/4/2000 1/4/2001 1/4/2002 1/4/2003 1/4/2004 1/4/2005 1/4/2006 1/4/2007 1/4/2008 1/4/2009 1/4/2010 1/4/2011 1/4/2012
Source: US Commodity Futures Trading Commission
Note: Total participation: commercial (black) and non-commercial (grey).
AGRICULTURE TRADING
251
Figure 10.2 Non-commercial trading as a percent of commercial participant

volumes for various agriculture markets
160.0%
140.0%
120.0%
100.0%
80.0%
60.0%
40.0%
20.0%
0.0%
rn
Le ttle
a
e
s
ga ilk
11
gs
So . 2
ns
e
s i il
l
Ro bo
at
ffe
co
ttl
ea
ric
ic
o
Co
ho
ea
o.
O
no
ca
ca
ju
c
co
Co
Ar n m
n
+k
rn
yb
ea
ii
an
ug
e
er
ve
a
ng
ot
yb
tto
ea
ic
ed
Li
as
cb
ra
ab
Su
So
yb
Co
Fe
Cl
O

So
at
he
W
thus being exempt from reporting. This occurs in all markets, but is
more pronounced in the livestock complex in general. The tradi-
tional commercials in live and feeder cattle are the feed yards, most
of which hedge their exposure in the live cattle. While cow/calf and
stocker operators utilise the feeder cattle market for hedging
purposes, the majority of their position sizes fall below the reporting
requirements.
Understanding the economics of physical commodity businesses
requires a strong knowledge of the individual components that
determine profit margins. This analysis of market fundamentals can
give traders an edge in generating opportunities and determining
the best types of trading strategy to implement. By understanding
the nuances of producer and merchant margins, non-commercial
traders can better assess buy-side and sell-side hedging activity that
takes place in the futures market. The most margin-sensitive hedgers
are active on both the buy- and sell-side; those include merchan-
disers, livestock feeders and processors. More traditional sell-side
hedgers include producers who have less market-related margin
risk, as their input costs are more tied to the operational overhead
and productivity. For instance, consider a grain farming operation:
in advance of each growing season, the producer must decide which
crop to plant by assessing a variety of important factors such as the
projected profitability per acre and the soil conditions across the
acreage in which the crop will be planted on. While the price of the
252
AGRICULTURE TRADING
underlying cash commodity is a driver of the producers margins, it

is not the sole influence of what a producer ultimately decides to
plant. The producer has to account for factors such as soil condition
and potential yield variability based on crop rotation practices that
can have important implications on the level of production per acre.
Equally important to profitability are overhead and input costs such
as seed, machinery, financing, labour and fertiliser. These factors
create a fixed piece of the margin that producers must account for in
advance of planting their crop; as a result, the selling or marketing of
that crop is a vital decision.
For non-commercial traders, understanding the economics behind
the sell-side hedgers decision-making can produce clear signals for
the future change in supply of a particular commodity. For example,
a noticeable lack of producer selling could indicate decreased
production for a commodity. In agriculture this could be due to a
poor growing season that has producers revising their expected
output, or it could be driven by the lack of economic incentive to
produce due to poor profit margins at the time of seeding.
Figure 10.3 highlights the growth in commercial trading across
individual markets. Note the growth in corn, sugar and soybeans, as
those commodities aside from traditional uses such as feed and
food have seen new demand come in the form of renewable energy
initiatives across the world. This relatively modern dynamic has had
both a direct and indirect impact across the agriculture market,
increasing participation by commercials and non-commercial
traders alike.
For example, the US Renewable Fuel Standard (RFS) requiring
gasoline refiners to blend corn ethanol was introduced in 2005. In
2007, the RFS mandate was increased to a 10% corn ethanol blend in
gasoline. The introduction and subsequent increase in the US renew-
able fuels mandate has resulted in increased demand and
competition for the US corn supply (see Figure 10.4). In 2011, around
40% of the domestic corn supply was consumed by the ethanol
industry. This additional demand has not only increased corn prices
but also that of competing row crops. As a result, the US RFS has had
a direct and meaningful impact on the US and global grain industry.
Consumers of grains have been affected as costs for feed and other
related inputs have increased in value. Markets such as livestock
have also been indirectly affected, as profit margins have at times
253
Figure 10.3 Commercial trading growth across individual agriculture markets
2,000,000
Corn
1,800,000
1,600,000
1,400,000
Contracts
1,200,000
Sugar No. 11
1,000,000
800,000
600,000
Soybeans
400,000
200,000
0
1/4/2000 7/4/2001 1/4/2003 7/4/2004 1/4/2006 7/4/2007 1/4/2009 7/4/2010 1/4/2012
Wheat CBOT Corn Soybeans Cotton No. 2 Lean Hogs Live Cattle Cocoa Sugar No. 11 Arabica Coffee
254
AGRICULTURE TRADING
been negatively impacted by higher corn values resulting in

producers decreasing herd size or seeking alternative feed rations.
Another indirect affect of the US RFS has been on the soybean meal
market; during the ethanol production process, a third of the caloric
value of corn is retained in a by product called distillers dried grains
(DDGs). The introduction and prominence of DDGS have presented
another source of feed for livestock and poultry producers that have
altered pricing relationships between soybean meal, hay and other
sources of protein and roughage.
Non-commercial traders
This section covers non-commercial traders by providing descrip-
tions of each type. This class of trading participant includes
fundamental discretionary and individuals trading proprietary
capital, to systematic and technical trading (all of which will be
detailed in this chapter). These traders can incorporate many
different forms of risk-taking based on return objectives, opportuni-
ties in their market and their approach to trading. Agriculture
markets present unique challenges and opportunities for non-
commercial traders due to risks involving seasonality, liquidity and
weather.
The fundamental discretionary trader uses fundamental analysis
Figure 10.4 Corn usage by segment, illustrating the importance of tracking

usage by end-users
7,000
Feed/residual FSI
6,000
5,000
4,000
3,000
Exports
2,000
1,000
Carry out
0
92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12*
Source: USDA ERS, Feed Outlook

* projection
255
to make trading decisions in the agriculture markets. Many of the

fundamental discretionary traders are registered with the US CFTC
as commodity trading advisors (CTAs), allowing them to market
themselves as an investment vehicle and manage client money in
individually separate managed accounts. There are also agriculture
specialist hedge funds that manage client money through onshore
and offshore vehicles. Since the beginning of the 21st century, the
agriculture markets have witnessed significant growth in the
number and size of assets under management and managers. The
increase in speculative trading across agriculture markets at the turn
of the century can be attributed to the evolution of electronic trading
as global speculators were increasingly allowed greater access, trans-
parency and flexibility to execute trades on commodity exchanges.
Inflationary risks have latterly attracted speculators, as global central
banks stimulus and US Federal Reserve policy measures have
increased the flow of money in the marketplace. Fundamentally
speaking, agriculture markets have been attractive in regard to theo-
ries and scientific research surrounding climate change and its
possible implications for the future of global agriculture production.
Additionally, social economics involving population growth,
changing dietary habits and emerging market demand have all had
an impact.
These traders commonly come from physical commodity back-
grounds for example, having worked as a grain merchandiser for
Cargill or a sugar trader at Louis Dreyfus. Other traders that have
built out money management businesses have come from the agri-
culture trading pits of Chicago, where they were successful
proprietary traders or brokers for large commodity customers. In
most cases, the fundamental discretionary commodity trader has
spent an invaluable portion of their career working for commodity
businesses, where they learned the fundamental pillars of what
drives supply and demand for each commodity they trade. Figure
10.5 illustrates the percentage of non-commercial trading relative to
commercial trading across agriculture markets since 2000.
Proprietary traders: individuals and trading groups

Proprietary traders are a diverse subset on their own, as this type of
trader fills in all the cracks inside the non-commercial participant
spectrum. The most common prop trader makes a living trading
256
Figure 10.5 Percent of non-commercial trading relative to that of commercial trading
100.00%
90.00%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
09
10
1
0
03
06
12
2
00
01
00
00
00
00
00
00
20
20
20
20
0
2
/2
/2
2
/2
/2
/2
/2
1/
1/
1/
1/
1/
1/
1/
1
01
1
01
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/
04
04
04
04
04
04
04
04
04
04
04
04
04
AGRICULTURE TRADING
257
their own capital. Historically, many of these traders operated on the

commodity exchanges in the trading pits as locals (a pit trader who
trades for themself), assuming 100% of their own trading risks. Over
the years, the number of proprietary trading firms (groups of propri-
etary traders within one organisation) has grown due to the rise in
electronic trading and also because of the profitability in profit
sharing that owning a proprietary trading group can have. This busi-
ness can be viewed as a private platform in which the owners of the
business hire talented individual traders, provide the overhead
including back office, administrative, accounting and trading tech-
nology for a share of any profits generated by the trader. Other
types of proprietary traders sometimes get unfairly described as less
knowledgeable or hot money. These are individuals who may not be
solely dependent on their success in trading commodities and at the
same time may not be aware of the significant risks that exist in
trading commodity futures. Both the type of trader and the amount
of capital traded is extremely diverse, from small accounts trading
under US$100k to multi-million dollar programs. This group of part-
time speculators participates in the same market as professional
traders, and sometimes has very different views of the commodity
they are trading. They may be prone to participate in crowded
or popular trades. In agriculture markets, proprietary traders
and trading groups provide significant daily liquidity for other
participants.
Systematic and technical traders

Systematic and technical traders, much like the proprietary trading
segment, are a vital part of the anatomy of the agriculture futures
and options markets as their trading volume provides commercials
and money managers essential liquidity that allows them to use
structured, fundamentally based strategies. Increased trading
volume can narrow bidoffer price spreads, allowing all trading
participants a better trade execution. In the systematic world, there
are very few money managers that trade purely in agriculture; many
of the commodity systematic programs will allocate a risk bucket
toward the agriculture markets in the range of 540% of their capital.
This is largely due to targeted capacity of assets under management
for the trading program relative to the capacity in the agriculture
markets. Additionally, factors such as style, strategy and correlations
258
AGRICULTURE TRADING
in respect to the systematic program models may dictate how much

the program allocates to agriculture markets. Pure discretionary
technical traders can be more opportunistic about their risk alloca-
tions across agriculture, which can provide outperformance relative
to other commodity sectors, resulting in an increased risk bucket.
Systematic programs trading in agriculture come in many
different forms, such as trend, multi-model, short-term momentum
and relative value. There has been considerable growth in systematic
programs which incorporate historical seasonality of prices and
spreads that have inherent fundamental ties. Some even will employ
econometric supply and demand modeling, which evaluate funda-
mental data produced for each commodity and then generates a
trading signal by way of a proprietary algorithm. This more sterile
and indirect fundamental trading from systematics can increase the
competitive advantage over discretionary participants due to the
discipline in generating and maintaining the trade. At other times,
this detachment can work against them as commodity fundamentals
can occasionally behave counter-seasonally and price patterns can
differ from historical norms which can give the advantage to the
discretionary manager who has the ability to adapt to the changing
environment. Counter-seasonal price behaviour can occur due to
supply/demand shocks. In turn, these shocks can be driven by
issues such as supply chain logistics, global trade flow and currency
valuations. On the macro side of things, geo-political and economic
risks can alter price behaviour.
High levels of adaptability can also be a characteristic of a talented
chart technician who trades breakouts and mean-reversion strategies
across the market. The chart technician relies on price data, behav-
iour and chart formations to produce trading signals, and
participates in price discovery and provides liquidity to the market.
Often, the discretionary technical and fundamental participants who
are into the right side of a breakout do so more quickly. For the
fundamental discretionary trader, this can be due to their funda-
mental analysis, while for the discretionary technician this can be
reactionary as their technical indicators (non-fundamental statistics
derived from the markets price data) signal them to enter a trade. On
the other hand, multi-model and trend-based systematic programs
will often be into a breakout or changing price environment only
after a trend in price can be confirmed.
259
Commodity index and swap trading

A passive and increasingly common form of trade flows comes from
commodity index fund and swap trading participation. A
commodity index is an index that tracks a basket of commodities to
measure their performance. Commodity indexes are often traded on
exchanges, allowing investors to gain easier access to commodities
without having to enter the futures markets. The value of these
indexes fluctuates based on their underlying commodities, and this
value can be traded on an exchange in much the same way as stock
index futures. There is a wide range of indexes on the market, each of
them varying by their components. The Dow Jones-UBS Commodity
Index (DJUBSCI), which is traded on the Chicago Mercantile
Exchange (CME), comprises 22 different commodities ranging from
aluminium to wheat. Index funds also vary in the way they are
weighted; some indexes, for instance, are equally weighted while
others have a predetermined, fixed weighting scheme. For example,
the DJUBSCI is reweighted and rebalanced annually on a price
percentage basis. While index fund trading flows are passive, they
have become more dynamic in their re-balancing and positioning
across the forward curves. Cleared commodity swap trading has
also become a larger piece of agriculture trading business by both
fundamental and speculative entities. A commodity swap is a
product whose exchanged cashflows are dependent on the price of
an underlying commodity. For commercial trading groups, a
commodity swap is usually used to hedge against the price of a
commodity. Therefore, in the case of a company that uses a lot of
corn, it might use a commodity swap to secure a maximum price for
oil. In return, the company receives payments based on the market
price. There are also cleared, over-the-counter (OTC) commodity
index swaps that allow investors to have direct exposure to a variety
of commodity or agriculture-specific indexes. Commodity index
swap contracts are based on indexes that are among those most
closely followed for investment performance in the commodity
markets. Investors, asset managers and financial institutions use
them to track performance or as benchmarks for their actively
managed accounts.
260
AGRICULTURE TRADING
TRADING IN AGRICULTURE MARKETS

Specialist traders in the agriculture sector use a wide range of non-
directional strategies, such as calendar/inter-commodity spreads,
and geographical and volatility focused arbitrage. The main drivers
of positive riskreward opportunities from non-directional strategies
come from the identification of possible structural shifts in the shape
of the forward price curve or term structure, and the expected
volatilities in between the spot month and deferred futures contracts.
By identifying mispricing relative to forecasted expectations
between differentials in terms of price and/or volatility, specialist
traders can structure dynamic non-directional strategies across the
forward price curve. Time horizons traded across agriculture tradi-
tionally have ranged from 13 months up to 612 months in order to
provide sufficient time in which a strategy can reflect a traders
supply/demand forecast.
However, given increased volatility and short-term spikes in
correlation driven by outside market influences, some more tradi-
tional intermediate to long-term discretionary fundamental traders
have adapted by ratcheting down their trade durations in response
to increased downside risks coupled with higher rates of return on
underlying strategies over short periods of time. Latterly, outside
market influences combined with increased speculative interest
across agriculture markets made more accessible by electronic
trading have resulted in short periods of high correlation across
markets. Traders and larger investment funds that manage a diverse
set of exposures can now more easily increase and decrease risk
across all markets in a more efficient and timely manner. In the event
of sudden geo-political or macroeconomic risks, these participants
can now enter and exit trades in a more concentrated fashion
causing cross-asset correlation to rise, typically only over short
periods of time (inside of one day to one week). At times where
prices skew non-fundamental due to a risk on/risk off environ-
ment, fundamental specialists are presented with the challenge of
appropriately managing risk while realising attractive trading
opportunities due to mispricing.
Market environments and volatility

The wider range in volatility across agriculture markets has
increased shorter-term trading by fundamental traders due to posi-
261
tive riskreward opportunities ie, allowing traders to avoid tying

up margin dollars for long periods of time while still allowing them
to continue trading a long-term theme. There are risks which make
shorter-term strategies more challenging, predicated on the trader
being able to quickly filter possible riskreward opportunities, all
while determining an appropriate size of risk allocation that is neces-
sary to achieve their profit target. Psychologically, this style of
trading requires steady and consistent discipline due to the limited
timeframe available to place the trade. Therefore, timing is critical in
order to have success in short trading frames. For traders aiming to
trade in and out of deferred contract months, narrow time horizons
can particularly be a challenge as pockets of less liquidity and wide
bidoffer spreads can cause slippage and dilute trading returns. For
example, a short-term trade in the 4th option of Kansas City Wheat
may look good on paper, but dried up liquidity as a result of a
pending crop report could cause wide bidoffer spreads, making it
difficult to implement or exit the strategy. In summary, most of the
difficulties in short-term trading are created by timing, lack of disci-
pline and varying degrees of liquidity.
Figures 10.6 and 10.7 illustrate the average true range (ATR) that is
a measure of volatility utilised by traders across the agriculture
space. Note the increased volatility in the ATR in both examples
shown.
Strategy selection
As volume and open interest vary across agriculture markets, so do
the type of suitable strategies and accompanying risks. Generally
speaking, total volumes and open interest in agriculture sub-sectors
rank in the following order, from largest to smallest: grain/oilseeds,
softs/tropicals and livestock/dairy. Given varying liquidity and
behaviour, traders must identify what strategies are best suited for
specific markets. This is especially the case for broadly diversified
commodity traders who may prefer taking a one-size-fits-all
approach to implementing and managing strategies across markets.
Specialist, individual market traders typically have a stronger handle
on risk tolerances and go-to strategies.
For example, relative value strategies in livestock that focus more
on pricing anomalies across the curve and less on absolute direction
work extremely well. While in the grains and oilseeds, more of a mix
262
Figure 10.6 Soybeans, weekly price and ATR
US$2,000.00 425.00
US$1,800.00 375.00
US$1,600.00 325.00
US$1,400.00
275.00
US$1,200.00
225.00
US$1,000.00
175.00
US$800.00
125.00
US$600.00
US$400.00 75.00
US$200.00 25.00
US$0.00 -25.00
9
01
05
09
1
97
07
98
99
99
99
00
01
19
20
20
20
19
20
1
/1
/1
/2
/2
6/
6/
6/
6/
6/
6/
6/
6/
6
06
6
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/
12
12
12
12
12
12
12
12
12
12
12
12
Soybeans cents/bushel Average true range, weekly
AGRICULTURE TRADING
Source: DTN ProphetX
263
12
/0
6 /1
0.00
5.00
10.00
15.00
20.00
25.00
30.00
35.00
40.00
9 89
12
/ 06
/1
9 91

12
/0
6 /1
99
3
12
/0
6 /1
99
5
12
/0
6 /1
99
7
Figure 10.7 No. 11 Sugar, weekly price and ATR
12
/0
6 /1
99
9
No.11 sugar, cents/pound
12
/0
6/
2 00
1
12
/ 06
/2
0 03
12
/0
6/
20
05
Average true range, weekly
12
/0
6/
20
07
12
/0
6/
20
09
12
/0
6/
20
11
0
5
10
2.5
7.5
1.25
3.75
6.25
8.75
264
AGRICULTURE TRADING
in options volatility, inter/intra commodity spreads along with flat

price strategies can offer better returns. Inter-spread is a cross-
commodity spread, in this case inside the grains and oilseeds sector
(for example, selling wheat and buying corn). Intra-spreads involve
spreads across the same commodity forward curve. Grains and
oilseeds offer traders a wide array of choices in terms of strategy util-
isation. The grains and oilseeds sector offer such a diverse and
attractive set of opportunities, such as inter-commodity relative
value that is, a spread between two commodities. Due to strong
competition for global production acres and substitutability factors
across grains and oilseeds products, traders like to implement strate-
gies that can exploit these fundamental relationships. Palm oil versus
canola oil or corn versus wheat are basic examples of global markets
that not only compete for production capacity, but for demand. The
fundamental competition inside the sector and the importance of
these markets globally is a strong reason why they offer relatively
deeper liquidity due to a more globally diverse set of participants.
Other sectors similar to livestock can be found in the tropical
commodity space, where sugar, coffee and cocoa specialists are
heavily reliant on managing relative value spreads and geographical
arbitrage. Table 10.1 outlines five types of trading strategies
commonly implemented across the agriculture commodities space.
Correlation benefit
Broadly diversified fundamental commodity traders have strong
incentives for including agriculture strategies in their portfolio, not
only because of stark fundamental differences and attractive themes
that exist across the sector. The diversity within the sector creates
significant de-correlation that does not always exist in other
commodity sectors, such as energy and metals. Correlations between
RBOB Gasoline, WTI Crude Oil, Brent Crude Oil or other energy
commodities can be high with each other, and they all tend to be
influenced by global macroeconomic headline risk and volatile stock
market fluctuations. Metals markets such as copper, aluminium, zinc
and palladium also show high correlations to each other. On the
other hand, across the agriculture markets one can find a number of
different combinations that offer low correlations for example, live
cattle versus sugar and cocoa versus corn, which helps create natural
diversification.
265
Table 10.1 Strategy types
Strategy types Description

#1 Directional Entering long or short futures and or options
across one or more contract months in one or
more commodities.
Example using futures Outright long December corn futures

Example using options Long October No. 11 Sugar 22 cent calls and
short 28 cent calls.
#2 Calendar spreads Simultaneously entering a L/S futures and or
options position across two different contract
months in the same underlying months in the
same underlying commodity market.
Example using futures Long March soybean futures and short July
soybean futures.
Example using options Long March soybean calls and long July soybean
puts.
#3 Geographical spread Simultaneously entering a long and short futures
arbitrage and/or options position across the same or
different contract months in two different
commodities.
Example using futures Long May Arabica coffee and short May Robusta
coffee.
#4 Crush spreads Simultaneously entering three legs in the futures
and/or options across three related commodities
by entering two buys and one sell, or two sells
and one buy. Often related to production margins
of a particular commodity.
Example using futures Soybean crush: Long soybeans, short soybean

meal, short soybean oil.
Example using futures Cattle crush: Long feeder cattle, long corn and
short live cattle.
#5 Options volatility Going L/S or spread commodities based on
implied and historical volatilities.
Example Relative value: Long December wheat calls at

25% volatility, short July wheat calls 40%
volatility.
Table 10.2 provides daily correlations across individual agricul-

ture commodities and comparative to energy and metals
commodities. The correlations in this table also show the distinct de-
266
Table 10.2 Daily correlations
Commodity CC KC SB FC LH LC C W S SM BO CT RR CL HO NG HG SI
Cocoa (CC)
Arabica coffee (KC) 72.86
No. 11 Sugar (SB) 69.13 47.75
Feeder cattle (FC) 63.37 29.38 54.30
Lean hogs (LH) 3.81 29.79 23.70 48.38
Live cattle (LC) 49.96 6.60 38.31 89.06 51.96
Corn (C) 25.84 46.29 0.49 16.58 63.64 30.97
Soft Red wheat (W) 47.38 23.57 32.39 43.55 9.07 32.20 57.21
Soybeans (S) 4.69 9.04 7.37 19.07 36.71 13.70 69.23 61.29
Soybean meal (SM) 14.21 36.58 23.40 20.08 23.46 9.78 50.74 55.22 94.10
Soybean oil (BO) 50.93 61.83 40.09 2.71 36.19 11.30 69.97 44.28 55.30 25.37
No. 2 cotton (CT) 83.91 72.27 53.60 51.97 0.69 37.89 23.22 41.66 0.66 20.02 55.57
Rough rice (RR) 0.87 19.63 12.57 20.24 32.41 30.09 44.74 6.38 33.36 22.20 34.07 25.90
WTI crude oil (CL) 2.08 31.11 17.89 43.36 30.28 49.29 38.81 10.17 10.66 5.29 50.28 25.00 9.47
Heating oil (HO) 11.22 33.30 22.32 67.18 61.20 76.27 59.52 15.80 23.82 5.90 52.91 0.38 35.96 81.16
Natural gas (NG) 79.97 72.04 61.03 74.30 1.03 59.59 16.69 41.39 12.04 28.39 33.90 69.12 1.17 15.42 25.36
AGRICULTURE TRADING
Copper (HG) 80.12 63.62 67.44 38.82 5.94 34.09 30.69 43.83 17.80 5.54 72.08 77.27 3.11 29.91 11.48 59.89
Silver (SI) 36.77 72.67 13.51 20.42 59.34 35.01 60.95 2.32 15.17 11.59 67.59 34.22 42.24 55.74 71.68 29.83 47.29
Gold (GC) 49.45 1.92 34.87 72.60 43.93 76.59 29.81 41.90 7.06 1.77 6.71 56.82 59.29 27.02 66.97 46.53 37.45 47.76
267
correlation across sectors such as tropical and livestock commodities

(ie, cocoa versus live cattle).
Figure 10.8 shows the correlation benefits across various pairings
of agriculture commodities such as corn versus feeder cattle.
Investment flows, seasonality and weather

Low correlations across agriculture commodities are driven by
market-specific supply/demand cycles, adverse weather and
seasonality, which can create a rich set of diverse trading opportuni-
ties. It should be noted that, with increased volumes and
participants, more traders are leaning on strategies tied to a variety
of historical seasonality features, making it increasingly challenging
to generate positive alpha. This has been witnessed in intra-
commodity relative value, which is individual commodity spreads.
For example, flat price seasonality on spot month lean hog and live
cattle markets has pronounced impacts on spreads between the
nearby and deferred prices across their respective forward curves.
With access to 30-plus years of historical futures spread data, more
and more traders are implementing spreads based on these strong
seasonal tendencies, thus at times diluting the riskreward profile
for spread trades relative to years passed. The popularity of seasonal
relative value trades has also increased mean reverting opportunities
for technical contrarians and fundamental specialists that are able to
identify if a spread has moved too far too fast.
The most successful traders are able to decipher the changing
influence of market participants, such as commercials, systematic
and swaps (as detailed in the first section of this chapter), and how
they impact seasonality and contribute to short- and long-term
cycles. For example, traditional or first-generation long-only swaps
managers are known to roll long positions from the fifth to the ninth
business day of the month; however, over time, the market response
to this practice by other speculative participants has caused swaps or
index funds to roll long positions earlier and later. In fact, index
funds have evolved their product suite, offering what are called
second- and third-generation products which adjust strategy for
curve contango or backwardation, attempting to capture alpha by
shifting their directional bets dynamically across the curve over opti-
mised time horizons. In this case, the product suite is the index
products created in addition to the conventional style index, such as
268
Figure 10.8 90-day rolling correlations
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
-10.00%
-20.00%
-30.00%
-40.00%
11
11
12
12
12
12
12
12
12
1
11
11
11
01
01
01
01
01
01
20
20
20
20
20
20
20
20
20
20
20
20
20
/2
/2
/2
7/
8/
9/
1/
2/
1/
2/
3/
4/
5/
6/
1/
3/
4/
5/
6/
0
2
/0
/0
/0
/0
/1
/1
/1
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
/0
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
Cocoa versus Coffee No.11 Sugar versus Cattle No.11 Sugar versus Corn Corn versus Feeder cattle
AGRICULTURE TRADING
269
the second- and third-generation products which incorporate

dynamic technical inputs such as open interest, volume and relative
performance across contract months to determine which contract
month to trade across the term structure. Traders who can filter and
understand the impact of important market factors such as weather,
demand and policy-related news (for example, changes in the regu-
latory environment) in an efficient manner will have a leg up on their
competition.
Figures 10.9 and 10.10 highlight the overall inconsistent correla-
tions between individual agriculture commodities such as wheat
and corn relative to gold and crude oil. Note that times of high corre-
lation for example, 90-day rolling correlation between corn and
crude oil +50% are often due to short-term periods of investment
flows driven by macroeconomic data and speculative re-balancing.
For discretionary agriculture traders, the main take away from
understanding index fund activity is that successful strategies must
withstand short-term periods of strong investment flows. Often,
price differentials across the forward curve of agriculture commodi-
ties can become skewed by such activity, by pushing prices out of
line with fundamental expectations. This type of price behaviour
caused by money flows often allows discretionary traders the oppor-
tunity to place complimentary spread strategies which are designed
to profit when the market corrects or reverts.
As previously mentioned, seasonal price behaviour can also
generate opportunities, as the underlying physical commodity
values react to productions cycles, weather events and seasonal
demand tendencies. These factors in normal environments have
created price activity that has produced consistent patterns over the
years (see Table 10.3 for more information about the planting of
grains and oilseeds). For example, corn and soybean volatility
seasonally strengthen during the US spring planting season and
peak during the growing season. In the lean hog market during the
late spring and early summer, increased demand for pork coupled
with a seasonal slowdown in production historically have supported
higher wholesale pork values and relatively higher lean hog futures
market prices in the summer contract months. Conversely, in the
autumn and winter, increased hog weights due to cooler tempera-
tures and cheaper/higher-quality feed create some of the best feed
conversions per animal units of the year, typically resulting in large
270
271
24
/ 01
/2
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
35.00%
40.00%
45.00%
01
1
24
/0
2/
20
11
24
/0
3 /2
0

11
24
/0
4 /2
0 11
24
/0
5/
20
11
24
/0
6 /2
01
1
24
/0
7 /2
0 11
24
/0
8/
20
11
24
/0
9/
20
11
24
Figure 10.9 90-day rolling correlation between wheat and gold
/1
0/
20
11
24
/1
1 /2
0 11
24
/1
Wheat versus gold
2 /2
0 11
24
/0
1 /2
0 12
24
/0
2 /2
0 12
24
/0
3 /2
0 12
24
/0
4/
20
12
24
/0
5 /2
0 12
24
/0
6 /2
0 12
24
/0
7 /2
0 12
AGRICULTURE TRADING

24
/0
1 /2
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
0 11
24
/0
2 /2
0 11
24
/0
3 /2
0

11
24
/0
4 /2
0 11
24
/0
5 /2
0 11
24
/0
6 /2
0 11
24
/0
7 /2
0 11
24
/0
8 /2
0 11
24
/0
9 /2
0 11
24
/1
0 /2
0 11
Figure 10.10 90-day rolling correlation between corn and crude oil
24
/1
1 /2
0 11
24
/1
Corn versus crude oil
2 /2
0 11
24
/0
1 /2
012
24
/0
2 /2
012
24
/0
3 /2
012
24
/0
4 /2
012
24
/0
5 /2
012
24
/0
6 /2
012
24
/0
7 /2
012
272
AGRICULTURE TRADING
market-ready supplies that drive cash and futures prices lower

during the autumn and winter contract months.
Figure 10.11 illustrates the seasonality in lean hog spreads. This
specific spread is of the February versus October contracts (same
calendar year). Note the bold black line (2012) has moved lower in
price earlier then previous years seasonal price action, albeit in the
same direction.
Figures 10.12 and 10.13 show the US Drought Monitor for the
beginning of the 2012 US summer growing season and near the end
of the US summer growing season. Note the beginning of the US
summer season was dry across much of the US Corn Belt but not in
drought (as illustrated in Figure 10.12), while by the end of the
summer most all of the US Corn Belt had fallen into severe drought
conditions.
Figure 10.14 shows the corresponding US crop conditions for corn
as the early season dryness evolved into a severe drought across the
US Corn Belt. Note the steep drop-off in conditions during the end of
June and throughout July 2012. Figure 10.15 illustrates the corre-
sponding response in corn prices as conditions were continually
downgraded during the US summer growing season thus
decreasing production forecasts.
Fundamental data points

When assessing agriculture markets, traders will often structure
strategies around specific data points or fundamental reports that are
issued for each commodity. In the US, official government funda-
mental supply/demand information is produced by the US
Department of Agriculture (USDA). The USDA has many domestic
field offices and divisions, along with foreign agriculture attachs
stationed in the worlds key agriculture producing regions. These
divisions are tasked with compiling, accounting and analysing
important data involving such things as cash grain receipts, whole-
sale and retail meat prices, survey results regarding prospective
plantings and on-farm grain stocks. Every month, the USDA releases
a world agriculture supply demand estimate (WASDE) that
produces global and domestic balance-sheet estimates for important
agriculture commodities. This is just one of the many fundamental
reports produced by the USDA. While some traders may not neces-
sarily trade off of supply and demand data for example, growing
273
Figure 10.11 Lean hog February versus October spread (10-year seasonal)-1.5250
22.5000
20.0000
17.5000
15.0000
13.5900
12.5000
2012
10.0000
2013
7.5000
4.9850 5.0000
3.3350
2.6750 2.5000
-0.1750
0
-0.9500
-1.5250
-2.4850 -2.5000
-3.1500
-5.0000
-7.5000 -7.5000
-10.0000
-12.5000
-14.1600
-15.0000
2008
-17.5000
Apr-12 May-12 Jun-12 Jul-12 Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13
274
AGRICULTURE TRADING
Figure 10.12 US Drought Monitor (May 29, 2012)
Source: National Drought Mitigation Center at the University of Nebraska-Lincoln
Figure 10.13 US Drought Monitor (August 14, 2012)
Source: National Drought Mitigation Center at the University of Nebraska-Lincoln
275
Figure 10.14 US crop progress and conditions: corn

USDA NASS
M April May June July August September October November
Good and excellent (percent)
80%
2010
70%
Condition year 2009
2008
2008 60%
2009 2011
2010 50%
2011
40%
2012
30%
2012
20%
100%
Nov 25
80%
Condition (percent)
Condition type
Excellent
60%
Good
Fair
Poor 40%
Very poor
20%
0%
100%
Doughing
80%
Emerged
Progress (percent)
Dented
60%
Silking
Mature
40%
Planted
Harvested
20%
0%
M April May June July August September October November
Progress year(s) 2012 2011 20072011
Source: National Agricultural Statistics Service (NASS), crop progress report
conditions in corn it is important to be cognisant of the release dates

of fundamental reports, as price volatility can fluctuate sharply post
the release of such information. For traders, an equally important
endeavour, aside from analysing the report information, is to filter
which sentiment indicators or reports best compliment their strategy
and style. From a risk management standpoint, traders can judge
sentiment more qualitatively by using their discretion regarding
certain data and market response. For example, a trader will grade
surveyed analyst expectations against actual reported information,
as this type of methodology can provide them with a strong read on
276
AGRICULTURE TRADING
Figure 10.15 Corn futures price, weekly line

US$850.00
US$800.00
US$750.00
Cents per bushel
US$700.00
US$650.00
!
!
US$600.00 !
US$550.00 !
2012: drought inspired rally
US$500.00
US$450.00
US$400.00
1/6/12 2/6/12 3/6/12 4/6/12 5/6/12 6/6/12 7/6/12 8/6/12
market sentiment which can, in turn, help in the management of risk

after the release of a fundamental report.
Global macro commodity managers will utilise macroeconomic
indicators as an overlay to trading in agriculture commodities, both
for risk management and portfolio/strategy structuring. In doing so,
some traders will build proprietary models or take advantage of
experience and intuition when assessing price activity in macro
markets such as stocks, US dollar index and energy and currency
markets. Fundamentally, the USDAs National Agriculture
Statistical Service and World Supply and Demand Forecasts produce
a wide range of agriculture research, surveys and periodic reports
for commodities such as corn, cotton, sugar and live cattle that
provide important information to the global market, offering guid-
ance for future supply/demand expectations. For example, in the
grain and oilseed commodities the USDA reports information on
stocks, seeded acres and growing conditions. It is also important for
agriculture traders to understand which commodities are most
consumer-sensitive or can be most susceptible to macroeconomic
risks. Commodities such as live cattle (beef), cotton and orange juice
can quickly reflect changes in retail demand.
Aside from tracking underlying cash and retail values of those
commodities, traders will also assess economic data in order to gain
an understanding of consumer sentiment, such as US monthly
employment data and the Consumer Confidence Index (CCI). On a
less-frequent basis, country-specific policy changes regarding such
things as global trade and renewable energy initiatives can have a
277
meaningful long-term impact on supply/demand and the global

trade of agriculture commodities. For example, in 2011 the US
entered into a bilateral free trade agreement with Colombia, which
came into effect in 2012. This comprehensive trade agreement elimi-
nates tariffs and other barriers to US exports, expands trade between
the two countries and promotes economic growth for both. The
International Trade Commission (ITC) has estimated that the tariff
reductions in the agreement will expand exports of US goods by
more than US$1.1 billion, supporting thousands of additional US
jobs. The ITC also projected that the agreement will increase US GDP
by US$2.5 billion.
Many agricultural commodities also will benefit, as more than half
of US farm exports to Colombia will become duty-free immediately,
and virtually all the remaining tariffs will be eliminated within 15
years. Colombia will immediately eliminate duties on wheat, barley,
soybeans, soybean meal and flour, high-quality beef, bacon, almost
all fruit and vegetable products, wheat, peanuts, whey, cotton and
the vast majority of processed products. The agreement also
provides duty-free tariff rate quotas (TRQ) on standard beef, chicken
leg quarters, dairy products, corn, sorghum, animal feeds, rice and
soybean oil. This is an example of a trade policy between two nations
that will have a long-term impact on prices and the supply chain of
some commodities.
Technical inputs
From a technical chart trading standpoint, agriculture markets
provide a good platform to trade a range of styles, including
breakout, mean reverting and trend following. Technical indicators
such as Fibonacci retracements, relative strength index (RSI), market
profile and a variety of moving averages are utilised by traders.
Studying open interest and volume as well as viewing charts across
different time horizons such as intra-day, daily, weekly and
monthly help put medium- to long-term strategies into perspec-
tive.
For fundamental discretionary traders, technical indicators do not
necessarily have to generate trade ideas, but rather provide a confir-
mation for the entry or exit of a strategy. An example would be a
trader who has an underlying bearish directional bias in a market
based on demand concerns, and at the same time recognises that the
278
AGRICULTURE TRADING
RSI indicator has fallen below the overbought level; just below this
price level, if the price weakness can be sustained, the price will be
able to drop below both the 20- and 100-day moving averages. This
confluence of signals can help confirm a potential entry point for the
bearish directional strategy. Using this methodology helps in adding
discipline, as it forces traders to adhere to the price action relative to
the technical signals, which can often indicate future longer-term
price movements before actual fundamental developments can be
realised. This is an important filter that can temper traders expecta-
tions behind their fundamental conviction about a commodity
market, and helps them to be patient in expressing strong convic-
tions. Overall, there are a variety of technical indicators that can be
used in assessing the agriculture markets and, most importantly,
they offer a non-biased overlay to discretionary decision-making.
Figure 10.16 illustrates a combination of technical indicators that
can be used to signal a trading opportunity. Note, the moving
average cross as the 20-day crosses over the 100-day to the downside.
Additionally, in advance of this cross the RSI had been testing over-
bought territory, which indicates that the market maybe reaching a
top. In the case of this illustration, this was true and the moving
average cross provided a confirmation and a sell signal. Figure 10.17
illustrates a combination of Fibonacci retracement and moving
average cross that can be used to signal a trading opportunity and
provide the trader with a back drop in which to balance expectations.
STRATEGIES AT PLAY IN THE AGRICULTURE MARKETS

The previous section provided a general description of the types,
behaviour and objectives of traders in the agriculture markets. This
section will categorise the specific types of strategies being employed
by those participants, along with their risks and management of such
strategies. There are five main strategy types covered: directional,
calendar spreads, geographical arbitrage, crush spreads and options
volatility. Methodologies used in trading strategies involve the
research and analysis of seasonality, forward curve structure and
fundamental factors. Using those factors, traders are then tasked
with choosing the most suitable strategy that aligns with their funda-
mental thesis or return objective.
279
Figure 10.16 Confluence of technical indicators signalling a trading opportunity
100.00 300.00
Price falling below both the 20- and 100-day moving averages
290.00
90.00 %
280.00
Relative strength index (RSI) indicating near overbought values
80.00 270.00
260.00
70.00 250.00
240.00
60.00
230.00
50.00 220.00
210.00
40.00
200.00
34.63
30.00 190.00
180.00
20.00 170.85
170.00
10.00 160.00
Arabica coffee, continuous daily bar 150.00

0.00 140.00
Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Jun-12 Jul-12 Aug-12
280
Figure 10.17 Use of Fibonacci retracement and moving average cross to identify a trading opportunity
2.40
November 2013 soybeans to December 2013 corn ratio
2.34 (100.0%) 2.35
2.30
Moving average cross
2.25
2.20 (61.8%)
2.20
2.16
2.16 (50.0%)
2.15
2.11 (38.2%)
2.10
2.07
2.06 (23.6%)
2.05
2.02
2.00
1.97 (0.0%)
1.95
100% retracement from highs
AGRICULTURE TRADING
Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Jun-12 Jul-12 Aug-12 Sep-12 Oct-12

281
Directional
Directional trading strategies are a very common style of trade
employed by both commercial and non-commercial trading partici-
pants. Commercial traders using this strategy will utilise flat price
trades to market or hedge production or commodity risk. In its
simplest form, this can be implemented as a flat price futures buy or
sell or as a hedge against an underlying physical commodity expo-
sure. For non-commercial traders, the flat price exposure is a source
of beta that compliments their speculative ideas on future price
direction. Flat price trades among the non-commercial and commer-
cial trading community can be expressed in many different forms.
Different style of directional bets include options spreads, risk rever-
sals such as owning a call and selling a put against the same
underlying contract month, and synthetic options that involve
trading futures and options in the same contract month.
Prior to entering a directional trade, traders must evaluate a
variety of riskreward factors such as selecting the appropriate
contract month across the forward curve and choosing the expected
time horizons for the trade, while also establishing risk allocation,
profit targets and stop/loss level(s). Experienced traders looking to
place a directional bet in an agriculture market are always aware of
the calendar as seasonality plays a large role in the risk profile of a
directional trade. After taking into account seasonal factors, the
trader will determine which contract month can best express their
ideas on fundamental price movements. Since many commodities
futures in the agriculture sector span multiple crop years, traders
have to make sure their fundamental thesis ties to the appropriate
time horizon in which they are trading.
For example, during the month of May, an oilseed trader becomes
bearish and decides to sell the US soybean market on expectations
for an above-average new crop production, but sells the old crop July
contract in order to express their bearishness; while being short is the
correct directional position, in this case it is not the correct contract
month or season to be short based on the fundamental thesis. This
trader is taking significant risk by holding a short position in an old
crop contract that may be trading off of different supply and demand
fundamentals. Additionally, the riskreward expectation for such a
trade could greatly underperform due to muted trade duration as
the July contract will have expired before new crop production is
282
AGRICULTURE TRADING
harvested, therefore never allowing the market to fully price in the

traders fundamental price forecast. See Table 10.3, as the global crop
timetable shows traditional planting and harvest time periods for
corn, soybeans and wheat produced around the world.
After determining the best point to be positioned, traders must
decide how much risk or what level of conviction they have in the
trade. This determination comes from a confluence of factors
involving the price forecast, market volatility and expectations for
trade duration. If a trader has strong confidence in their fundamental
thesis and long-term price forecast, but the market volatility is high
due to shorter-term factors, they may take a scale in approach to
their directional position. Scaling into a strategy is a methodology in
which a trader increases risk by adding positions to the existing
strategy. This allows them to ultimately reach a high conviction or
risk allocation, while withstanding near-term volatility pressures.
Regardless of the conservative approach, traders still need to deter-
mine levels in which they will stop out of the directional position and
go to the sidelines. While liquidity in executing directional trades is
often better than liquidity available for more complex relative value
strategies, the returns on an unhedged directional trade can often be
more volatile, which makes risk management and position sizing
important.
Figure 10.18 offers an example of this type of risk differential,
showing the difference in ATR between an old/new crop corn
spread versus the individual components of the spread. Note the
ATR of the individual components, in this case July 2012 and
December 2012 corn traded as much as two or three times more on a
daily basis than that of the JulyDecember 2012 calendar spread.
Additionally, note the convergence and divergence of the spread
relative to the individual components, as the tug of war between old
and new crop supply/demand played out over time.
Calendar spreads
Calendar spread strategies have grown in popularity among the
speculative trading community due to their embedded alpha gener-
ation and strong relationship with fundamental price relationships.
As defined in Table 10.3, a calendar spread trade is a strategy in
which a buy and sell are simultaneously placed across the same
commodity futures curve. Calendar spreads provide fundamental
283
284

Table 10.3 Global crop timetable and futures contracts
Futures contracts symbols F G H J K M N Q U V X Z
Wheat Jan Feb March Apr May June July Aug Sep Oct Nov Dec
US Winter Harvests Plants
Soft Red Winter (W) WH WK WN WU WZ
Hard Red Winter (KW) KWH KWK KWN KWU KWZ
US Spring Plants Harvests
Hard Red Spring (MW) MWH MWK MWN MWU MWZ
Canada Plants Harvests
France Harvests Plants
Milling Wheat (PM) PMF PMH PMK PMQ PMX

Germany Harvests Plants
UK Harvests Plants
Ukraine Harvests Plants
Turkey Harvests Plants
Egypt Harvests Plants
Kazakhistan Harvests Plants
Russia Winter Harvests Plants
Russia Spring Plants Harvests
Iran Harvests Plants
Pakistan Harvests Plants
India Harvests Plants
China Harvests Plants

Australia Plants Harvests
Brazil Plants Harvests
Argentina Plants Harvests
Soybeans Jan Feb March Apr May June July Aug Sep Oct Nov Dec
Brazil Harvests Plants
Argentina Harvests Plants
US Plants Harvests
Soybeans (S) SF SH SK SN SQ SX
China Plants Harvests
Corn Jan Feb March Apr May June July Aug Sep Oct Nov Dec
Brazil Harvests Plants
Argentina Harvests Plants
US Plants Harvests
Corn (C) CH CK CN CU CZ
China (North) Plants Harvests
China (South) Plants Harvests
France Plants Harvests
Spain Plants Harvests
AGRICULTURE TRADING
Ukraine Plants Harvests
Russia Plants
India Plants Harvests
South Africa Harvests Plants

285
24
/0 Cents/bushel
5/
0
5
10
15
20
25
20
24 1 0
/0
6/2
01
24
/0 0
7/
20
24 10
/0

8/
20
24 10
/0
9/
20
24 10
/1
0/
20
24 10
/1
July 2012 Corn (Left Axis)

1/
20
24 10
/1
2/
20
24 10
/0
1/
20
24 11
/0
2/
20
24 11
/0
3/
20
24 11
/0
4/
20
24 11
/0
5/
20
11
December 2012 Corn (Left Axis)
24
/0
6/
20
24 11
/0
7/
20
24 11
/0
8/
20
24 11
/0
Figure 10.18 20-day ATR: old versus new crop corn spread relative to outright contracts
9/
20
24 11
/1
0/
20
24 11
/1
1/
20
24 11
/1
2/
20
24 11
/0
1/
20
24 12
/0
2/
20
July-December 2012 Corn Spread (Right Axis)
24 12
/0
3/
20
12
0
1
2
3
4
5
6
7
286
AGRICULTURE TRADING
traders with a non-directional bias and the opportunity to trade rela-

tive fundamentals across the term structure of a commodity.
Non-directional reasons to trade calendar spreads can involve price
relationships regarding cash basis and seasonality.
From a directional standpoint, some traders may entertain trading
a calendar spread as a hedge against being directionally positioned
at different points on the futures curve or as a more conservative bet
on directional price expectations against one leg of the spread. There
are many possible fundamental and technical drivers for trading
calendar spreads. Some of the most compelling calendar spread
strategies can be seen in Table 10.4.
Geographical spread arbitrage

Geographical arbitrage is another form of inter-commodity spread in
which a trader buys and sells the same type of commodity produced
across different regions of the world. These commodity futures
contracts often exist on different exchanges and have different
quality or grade characteristics. An example of trading a geograph-
ical spread would be purchasing Arabica coffee and selling Robusta
coffee. Trading a geographical arbitrage strategy is mainly carried
out by fundamental specialists due to the high level of specific
knowledge needed to understand the pricing relationships. For tech-
nical traders, this type of commodity spread can have appeal from a
mean reverting standpoint, as the trader will seek opportunities
when the spread between the two related commodities reaches
extreme levels. Purely trading geographical arbitrage from a tech-
nical standpoint, however, does come with significant risk as the flat
price direction of an individual leg of the spread can move oppo-
sitely for sustained periods of time based on specific
micro-fundamental factors. Other inter-commodity spreads can
have strong quality-based and seasonal aspects, such as trading
lower-grade US soft red winter wheat versus higher-grade hard red
spring wheat.
See Table 10.5, which details fundamental drivers for trading
inter-commodity or geographical arbitrage. The same technical
drivers can apply for these types of spreads as that outlined for
calendar spreads.
Figure 10.19 shows how the soybean to corn ratio provides an
example of blending technical, seasonals and fundamentals while
287
Table 10.4 Strategy types: calendar spreads
Fundamental #1 Old crop S&D versus new crop S&D forecasts.

reasons
#2 Individual flat price biases: bullish and bearish across
different time horizons.
#3 Seasonality of basis (cash minus futures) versus
forecasted basis.
#4 End-user and producer profit margins impact on
underlying cash values.
Technical reasons #1 Seasonality of the spread differential.
#2 Bull or bear spread as a hedge against directional bias.
#3 Bull or bear spread as a theoretical conservative bet on
directonal bias.
#4 Commitment of traders data.
Geographical Simultaneously entering a long and short futures and or
spread arbitrage options position across the same or different contract
months in two different commodities.
Example using Long May Arabica coffee and short May Robusta coffee.
futures
Crush spreads Simultaneously entering three legs in the futures and or
options across three related commodities by entering two
buys and one sell, or two sells and one buy. Often related
to production margins of a particular commodity.
Example using Soybean crush: long soybeans, short soybean meal, short
futures soybean oil.
Example using Cattle crush: long feeder cattle, long corn and short live
futures cattle.
Options volatility Going L/S or spread commodities based on implied and
historical volatilities.
Example Relative value: long December wheat calls at 25%

volatility, short July wheat calls 40% volatility.
assessing a possible geographical arbitrage spread opportunity. The

eight-year seasonals show the behaviour of the ratio to be rather
inconsistent, but do provide a range of expectations. It is up to the
trader to deduce what fundamental drivers will result in the future
performance of such types of geographical arbitrage, as each year
can be extremely different from the next.
288
AGRICULTURE TRADING
Table 10.5 Strategy types: geographic arbitrage
Fundamental #1 Trading differences in supply and demand across

reasons different regions.
#2 Trading differences in quality grades of a similar
commodity.
#3 Trading differences in localised demand and its impact
on underlying cash prices.
#4 Trading the flow of a similar or competing commodity
based on supply, demand and logistics.
Technical reasons #1 Seasonality of the spread differential.
directonal bias.
spread arbitrage options position across the same or different contract in
two different commodities.
futures
futures soybean oil
futures cattle

Crush spreads
A crush spread is a form of arbitrage predominately used by
commercial traders in order to manage production-related margin
risk. Typically, a crush spread includes two or three individual
components. Speculative participants with a keen understanding of
production margins often like to implement crush or reverse crush
spreads as a proxy as it allows them to participate synthetically in
289
Figure 10.19 March 2013 soybeans to corn ratio (eight-year seasonal)
3.40
3.23
3.20
3.01
3.00
2.80
2.62
2.60
2.51
2.46
2.40
2.35
2.25
2.20
2.09
2.02
2.00
1.89
1.80
2013 ratio bold black line
1.60
Dec-11 Jan-12 Feb-12 Mar-12 Apr-12 May-12 Jun-12 Jul-12 Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13
290
AGRICULTURE TRADING
physical commodity margins. A good example of a crush spread in

agriculture can be found in soybeans, where a trader can replicate a
physical soybean crushing plant, by purchasing soybeans and selling
the output including soybean meal and soybean oil contracts. Other
agriculture commodities in which crush trading is popular are live-
stock, where producers in the pork and beef industries will actively
Table 10.6 Strategy types: crush spreads
Fundamental #1 Trading the production economics or margins of a

reasons specific commodity.
#2 Trading differences in margins of a particular
commodity across the forward curve via calendar crush
spreads.
#3 Trading the reverse crush by taking the opposing side of
the relationship typically held by the physical commodity
producer.
Technical reasons #1 Seasonaility of the spread differential.
directonal bias.
spread arbitrage options position across the same or different contract in
two different commodities.
futures
futures cattle.

291
purchase soybean meal and corn against the lean hog or live cattle
futures. Table 10.6 details the fundamental drivers for placing a
crush trade.
Options volatility
Trading of options volatility strategies offers traders with a wide
range of dynamic opportunities on a standalone basis, and also when
coupled with futures directional and relative value spreads. Trading
opportunities in options include individual commodity spreads and
direction or across commodities in the form of arbitrage.
Experienced relative value option specialists in agriculture are
frequently able to find attractive opportunities by trading differen-
tials in volatility on an inter/intra commodity basis. Additional
strategies involve trading put versus call skews across one or more
contract months in one or multiple commodities.
Directional trading is also prominent in options by way of owning
net, absolute gamma or premium in any contract month. An example
of a net gamma options play would be to own a bull call spread in
which the trader purchases an at-the-money call and sells an out-of-
the money call against it at a slightly lesser value, resulting in a net
payment of premium and a net long volatility position. The number
of options strategies which can be expressed across agriculture
markets is seemingly endless, and they provide traders with unique
and niche opportunities to generate profitable returns. Table 10.7
details the three different types of options strategies that are often
traded across the agriculture space.
CONCLUSION
The speed of information flow and the sudden correlations across
markets from time to time can in some ways be attributed to the
success and growth of electronic trading, as a more diverse set of
speculative participants from around the world have virtual around
the clock access to trade and manage risk in most commodity
markets. In general, this new normal in price behaviour and
volatility offers more opportunities for multi-strategy and relative
value driven traders. Periods of high volatility and relatively wider
price ranges can frequently distort prices relative to perceived funda-
mentals, which can create unique opportunities. These types of price
environments are often associated with adverse market conditions;
292
AGRICULTURE TRADING
Table 10.7 Strategy types: options
Price distribution Trading the difference or skew in the options pricing

against same underlying futures contract.
This can be done by trading straddles or strangles based on
options pricing differentials.
Relative volatility Trading statistical differences in volatility between
correlated and or non-correlated commodities.
This can be done by selling relatively high volatility in one
commodity and purchasing relatively cheap volatility in
another.
Trading the difference between implied volatility and
historical volatility in one commodity.
This can be done by buying or selling volatility in one
commodity based on the relationship between implied and
historical volatility.
Relative value Trading the price relationship of an underlying futures
spread by way of using options.
This can be done by trading put, call and butterfly options
spreads on an inter/intra commodity basis.
futures
futures cattle.

those traders which can realise the difference between an event that
is normally anticipated (seasonal or fundamental data point) and one
that is rare and unexpected will find success in trading and
managing risk in agriculture markets.
The ability to recognise, filter, and accurately assess changing
market developments is critical in making trading decisions. With
293
the globalisation of agribusiness and trade expected to grow, so will

the expansion and enhancement of global agriculture futures and
options markets, which will further increase the set of trading oppor-
tunities available to all traders.
294
11
Quantitative Approaches to Capturing
Commodity Risk Premiums
Mark Hooker and Paul Lucek
State Street Global Advisors and SSARIS Advisors
Many institutional investors now allocate to commodities alongside

the traditional asset classes of equities and fixed income, based on
the primary motivations of diversification and protection against the
risk of inflation. While commodities have indeed been less correlated
to equities and many other risky assets (and offered comparable
riskreturn trade-offs, see Gorton and Rouwenhorst, 2008), in this
chapter we will show that those diversification benefits may be
enhanced through a deeper understanding of how and why different
passive and active strategies tend to perform in different market
environments. Our conceptual frame of reference views most invest-
ment strategies as either convergent or divergent performing well
in either normal or more dislocated periods and is applicable at
any level of aggregation, from individual securities to sectors and
markets as well as combinations of asset classes.
We will begin with a brief review of commodity benchmarks,
highlighting the degree to which they are considerably less passive
than traditional equity and fixed income benchmarks, as well as an
overview of active approaches to commodity investing. We will then
present a detailed discussion of the convergent/divergent paradigm,
and demonstrate with an example how it can be applied within an
active commodities strategy, underscoring its effectiveness during
the most recent global financial crisis when traditional approaches to
diversification largely failed.
295
REVIEW OF COMMODITY BENCHMARKS AND OVERVIEW OF

ACTIVE APPROACHES TO COMMODITY INVESTING
Commodity benchmarks differ from their stock and bond counter-
parts in two main ways. First, there is no straightforward analogue to
market capitalisation for determining component weights eg, the
index weight of crude oil relative to that of soymeal. Instead, various
commodity indexes, including the prominent DJ-UBS and GSCI
indexes, use factors such as trading volumes of the futures contracts
and worldwide production statistics to derive individual component
weights. The collection of production and trading volume data is
also subject to a series of decisions regarding sources, timing, data
revisions and additional factors. Commodity benchmarks are there-
fore subject to a much greater degree of subjectivity in determining
benchmark weights. Second, since commodity futures contracts
have a limited time before their expiration, a set of rules must be
constructed to determine when contracts are rolled from the near
month to a later-dated month. These rules must indicate whether
adjacent contract months are used, or if certain months are skipped.
In order to minimise the impact of the change over from one contract
to the next, the roll usually occurs over a period of several days,
which also must be specified within the rules. In this sense, passive
commodity investing should be considered semi-active.
Active commodities strategies

The broad universe of commodity futures contracts exhibits a very
low average correlation of its components, a wide dispersion of indi-
vidual commodity futures returns, high volatility and large
drawdowns. For example, pairwise correlations between industry
group returns in the MSCI World Equity Index average about 0.5,
while analogous correlations between the constituents of the DJ-UBS
commodity index are closer to 0.2, and volatilities average about 50%
greater for commodities, at 30% versus 20% for equities.1 This
volatility and dispersion provides considerable opportunity for
skilled active managers to implement strategies that can provide
alpha over commodity index beta, while using risk controls to
reduce volatility and preserve capital. The combination of these
opportunities for active managers, in conjunction with the semi-
active nature of the commodity index providers, makes a strong case
for active commodity management within an institutional portfolio.
296
QUANTITATIVE APPROACHES TO CAPTURING COMMODITY RISK PREMIUMS
Strategies for active commodity management generally fall within

two categories: discretionary and systematic. In this chapter, we will
focus on systematic strategies, which are quantitative in nature,
using historical data to develop models that drive trading decisions
directly from feeding live data through the models. Quantitative
trading strategies are of course sensitive to the performance of their
underlying models, which typically have positive periods where
they perform as designed, and negative periods where factors
external to those included within the model drive more of the market
movements. For this reason, diversification of differently designed
trading models and approaches can greatly enhance the overall effi-
ciency of a quantitatively managed portfolio of commodity futures.
CONVERGENT AND DIVERGENT STRATEGIES

The convergent/divergent paradigm was introduced in Chung,
Rosenberg and Tomeo (2004). It focuses on distribution of monthly
returns, their statistical properties and the cross-correlations
between those returns and aspects of the market environment.
Convergent return streams have monthly return distributions
represented by the shaded curve in Figure 11.1, and are generally
derived from fundamental or value-based methodologies. In a
convergent strategy, a manager often calculates an intrinsic or fair
value for an asset: a target price. If the asset is trading below the
intrinsic target price, the manager would seek to buy the under-
valued asset. Conversely, if the asset is trading above the intrinsic
target price, the manager would seek to sell the overvalued asset.
The goal of the traded position would be for the current asset price to
converge to the target price and generate a positive return. The
manager seeks over- or undervalued assets with the expectation that
these assets will move toward their fair values, allowing them to
exploit the temporary mispricing. Convergent strategies tend to be
based upon fundamental methodologies, although certain quantita-
tive methods for example, mean reversion strategies also tend to
produce convergent return streams.
Most passive investments indexes are also convergent in their
nature. Passive index investing has the goal of capturing the risk
premium of the asset class. Fixed income risk premiums come from
credit and duration risks. The equity risk premium is associated with
earnings growth. Commodity risk premiums are derived from the
297
Figure 11.1 Convergent and divergent monthly return distributions
CONVERGENT
DIVERGENT
CONVERGENT DIVERGENT
| | | |
-2 0 2 5
Source: SSARIS/SSgA
inventory levels of the underlying commodities (Gorton, Hayashi

and Rouwenhorst, 2006). A passive investor in these indexes is
looking for the index return to converge to the expected risk
premium. Furthermore, as emphasised in Ilmanen (2011), asset class
risk premiums tend to be larger for investments that perform
poorly during crisis periods (bad times) so that they have some
characteristics of selling insurance. Assets that produce positive
returns in normal periods and suffer large losses in crisis periods are
convergent.
Convergent investments normally exhibit fairly consistent return
streams with a high frequency of small positive returns. Their consis-
tency and low volatility can give them high Sharpe ratios and make
these types of strategies very attractive to investors. One of the weak-
nesses that convergent strategies exhibit is their negative skewness.
As shown in Figure 11.1, the convergent return distribution has a
significant left-hand fat tail. After a series of several monthly returns
clumped around zero with a positive mean, market events can occur
where convergent strategies experience significantly (23 standard
deviation or more) negative returns. These events tend to be termed
crisis events, such as the 1987 stock market crash, the 1997/98
Asian currency/Russian debt/LTCM crises, the 200708 global
298
financial crisis (GFC) and the 2011 European debt crisis. During these
crisis events, fundamental and value-based strategies often have
significantly negative performance. When behavioural finance
concepts such as fear and greed drive market movements, asset
prices succumb to panic and overshoot their fair values. Convergent
approaches have great difficulty in this type of crisis environment
because, as an asset price drops due to panic and fear, the convergent
model suggests that the asset is an even more attractive buy. The
model will eventually be correct when the asset price hits a bottom
and the crisis passes, but trading positions taken along the way may
experience heavy losses. Unrealised losses in commodity futures
contracts force future commission merchants (FCMs) to issue margin
calls. If further capital is not produced, the managers positions are
liquidated and the losses are realised. This situation was aptly
described by a quote attributed to John Maynard Keynes: Markets
can remain irrational longer than you can remain solvent.
A prime example of crisis price dynamics is illustrated in Figure
11.2: the price of the December 2008 Nymex crude oil futures
contract. Within a span of 10 months, the contract rose from
US$84.62 per barrel to US$146.68, before sinking to US$49.62.
Somewhere within the range of a 73% run-up and a 66% decline was
an intrinsic value for crude oil, but the price had been driven far
beyond fair value in both directions.
During market dislocations, such large directional moves are
common. The most striking characteristic of these crisis events is an
increase in market volatility (almost, by definition, a crisis event
includes an increase in market volatility). A secondary effect is an
increase in magnitudes of correlations. Assets that previously exhib-
ited low correlations tend to become correlated during a crisis. A
tertiary effect is the increase in serial price correlation or autocorrela-
tion within individual markets. Table 11.1 shows these three effects
during the 200708 GFC: over 2007, the DJ-UBS index had an annu-
alised volatility of 12%, average correlation of its components of 0.15
and near-zero autocorrelation of those components returns. During
2008, each of those statistics more than doubled, with serial correla-
tion rising more than five-fold. When markets become driven
beyond fair value and fundamental convergent methodologies fail, it
is the divergent category of strategies that can capitalise on the
increase in market autocorrelation.
299
Figure 11.2 WTI crude oil contract (December 2008)
145
125
US$/barrel
105
85
65
45
16 /08
30 /08
13 /08
27 /08
12 /08
26 /08
9/ 08
23 08
7/ 08
21 /08
4 / 08
18 /08
2/ 08
16 /08
30 /08
13 /08
27 /08
10 /08
24 /08
8/ /08
/1 8
5/ /08
/1 8
08
22 0/0
19 /0
/
4/
1/
1
/1
/1
/2
/2
/3
/3
/4
5
/5
6
/6
7
/7
/7
/8
/8
/9
/9
0
11
2/
1
Source: Commodity Systems Inc
Divergent strategies capitalise on directional market moves. These

strategies, which include momentum and other trend-following
approaches, seek to take positions based upon the analysis of histor-
ical price data and the direction the market is currently moving. They
tend to perform well as the rate of change in volatility levels
increases, which is also when markets tend to exhibit more
pronounced degrees of autocorrelation.
While in normal or rational market environments convergent and
divergent strategies are usually uncorrelated, during crisis events
and irrational market environments the two become negatively
correlated. Divergent strategies perform well and experience right-
hand tail events at the same time that convergent strategies have
their negative left-hand tail events. When markets exhibit strong
directional moves such as with crude oil in 2008, momentum/trend
strategies can capitalise on the shifts away from fair value. Divergent
strategies in this sense are directly opposite to convergent strategies.
As crude oil rose in 2008 and became more and more expensive,
300
Table 11.1 DJ-UBS index trailing 52-week statistics, weekly data points
Annualised Average Average

standard correlation of auto-correlation
deviation components within components
2/1/2008 0.120 0.150 0.025

31/12/2008 0.293 0.429 0.145
increase 2.44 2.87 5.81
Source: DJ-UBS index
convergent models saw the asset as overpriced. Divergent models

saw the increasing price as a trend and favoured the asset. Divergent
strategies require a significant retracement in market price before
they will change their assessment of a market trend. Similarly,
during the market decline in the second half of 2008, as crude oil
became less and less expensive, convergent models favoured the
asset, while the strong downward move forced divergent trend
models to sell the asset.
The contrasting styles of convergent and divergent strategies and
the diversification of their respective return stream distributions
leads to the benefit of allocating to both strategies. Commodity
markets facilitate this diversification, because with the wide disper-
sion and low average correlation within the commodity universe,
divergent events can occur in one commodity sector while other
sectors exhibit a largely convergent environment. For example, in
2012 significant bullish moves in agricultural markets took place due
to the drought conditions in the US, with corn up more than 60% and
soybeans, meal and oil up roughly 25% between mid-June and early
August. These markets exhibited strong directional trends that
divergent strategies were able to capture, while other market sectors
provided strong performance from convergent strategies.
Investment strategies that allocate to successful convergent and
divergent techniques will outperform those that allocate to only one
or the other on a risk-adjusted basis.
A CONVERGENT/DIVERGENT ACTIVE COMMODITIES EXAMPLE

In order to demonstrate the beneficial effect of a convergent/diver-
gent quantitative approach to commodity investing, we present here
examples of the two trading methodologies, their individual perfor-
mance relative to a passive benchmark index and the increased
301
efficiency achieved by their combination. The sample period for this

analysis is January 1996 to October 2012.
A simple divergent momentum strategy described by Spurgin
(1999) uses three lookback timeframes in order to determine long or
short positions in each market traded. The three timeframes 15, 27
and 55 days were selected to best replicate an index of commodity
trading advisor (CTA) managers trading a broad range of futures
contracts. The signal derived from the momentum strategy is gener-
ated by comparing the price at time t with the price at a fixed number
of days ago. For example, in the 15-day system, if Pt > Pt-15, the
market trend is considered to be positive and a long position is taken.
If Pt<Pt-15, the market direction is considered to be negative and a
short position is taken. If the prices are the same, no market direction
is indicated and no position is taken. Trading signals from the three
lookback timeframes are averaged. In this chapter, we modify the
momentum trading methodology by not taking any net short posi-
tions. If an aggregate signal is negative, indicating a short, a flat
position is taken instead. This modification has the effect of making
the momentum system track the index more closely, since the index
itself does not take short positions. Signals for each of the 20 compo-
nents of the DJ-UBS index are generated on a daily basis. Profits and
losses were calculated by equally weighting the systems return
streams on each of the 20 components and rebalancing on a monthly
basis. The DJ-UBS index is presented as a benchmark for comparison
in Table 11.2. The momentum strategy outperforms the DJ-UBS
index while exhibiting less than half of its volatility.
For convergent strategies, we investigate two quantitative
approaches. The first relates to the GortonHayashiRouwenhorst
view that inventories are the key fundamental for commodity
futures returns, and that the degree of contango or backwardation in
the futures curve reflects those fundamentals. Here, we construct an
alpha signal for each commodity each month. That alpha is a simple
function of the relationship between the second-from-expiration
futures price and the next-to-expiration contract price.
The second convergent approach models each commoditys
returns using a Markov switching model. Here, each commoditys
monthly return is assumed to be drawn from one of three different
normal distributions, called regimes, each with its own mean and
variance. The process moves from one distribution to another (a
302
Table 11.2 Performance of divergent and convergent strategies versus the DJ-UBS
DJ-UBS Divergent Convergent Combined

commodity index momentum backwardation & convergent &
total return strategy (15-27-55 regime switching divergent strategy
day lookbacks) strategy (50%/50%)*
Annualised
return 0.048 0.053 0.106 0.083
Annualised
standard
deviation 0.167 0.077 0.203 0.132
Return/risk
ratio 0.289 0.681 0.523 0.626
Data based upon DJ-UBS Commodity Index returns (January 1996October 2012)
* The combined convergent/divergent strategy is rebalanced monthly
different normal distribution contributing the return draw) according

to a set of Markov transition probabilities. Expected returns are
computed each month for each commodity by multiplying the esti-
mated probabilities of being in each regime by their respective means.
Those means, variances and transition probabilities are estimated
periodically through time and evolve as new data accumulates.
The alpha estimates from these two convergent approaches are
averaged, and then run through a mean-variance optimisation to
produce over- and under-weights relative to the DJ-UBS Commodity
Index benchmark weights that also satisfy various constraints, such
as limits on short positions and on active exposures to individual
commodities and commodity groups. The return stream produced is
summarised in the final column of Table 11.2. The convergent
strategy also outperformed the DJ-UBS index, but with about a third
greater volatility, for a return-to-risk ratio somewhere in between the
index and the divergent strategy.
As mentioned above, convergent strategies tend to perform well
when assets move toward intrinsic or fair valuations, which tends to
be when markets are in more normal regimes or recovering toward
normality after severe shocks. By contrast, divergent strategies tend
to perform well during crises when markets are driven beyond fair
values. Consistent with this logic, the convergent backwardation
Markov switching strategy had average excess returns that were
slightly negative in 2008 as markets reacted severely to the global
financial crisis, but outperformed by more than 17 percentage points
303
in 2009 as markets dramatically recovered. In contrast, the divergent

momentum strategy outperformed the index by more than 45% in
2008, and underperformed by more than 12% in 2009.
The most striking aspect of the convergent and divergent excess
return streams is how their correlation varies across time. The overall
correlation is slightly negative, at 0.16. However, during the crisis
and recovery period of 200809, it rose dramatically (in absolute
terms) to 0.51, demonstrating the very powerful benefit through the
diversification achieved by combining these two return streams. A
programme allocating one half of the portfolio to each strategy and
rebalanced on a monthly basis showed an 8.3% annualised rate of
return with a 13% annualised standard deviation (outperforming the
benchmark index by more than 3%, with volatility reduced by more
than 3%) in a backtest over the test period of January 1996October
2012.
CONCLUSION
Most approaches to diversification focus on average correlations
over periods of time that encompass full market cycles indeed, that
is often by design, implicitly assuming that fluctuations in correla-
tions through time are primarily noise and so should be averaged
out. That approach has disappointed investors during crises as
realised correlations, particularly within the dominant convergent
set, all go toward one. The convergent/divergent paradigm, by
contrast, views variations in correlations particularly as a function
of market stress and stability as fundamental attributes of a
strategy that should be incorporated into portfolio design.
The example in this chapter showed that combining convergent
and divergent active quantitative strategies can provide significant
alpha over a passive index and stabilise the overall return stream of
the portfolio, especially during crisis event periods. In practice, we
have seen this concept can apply to fundamental active strategies,
passive strategies, other asset classes and the overall portfolio level
as well, where it may help overcome the challenge that diversifica-
tion has often worked least well when it has been needed most.
1 There are 23 industry groups in the MSCI GICS system versus 20 commodities in DJ-UBS, so
this level of aggregation makes them reasonably comparable. Correlations and volatilities
are computed using monthly returns from 1999 to 2009.
304
REFERENCES
Chung, S., M. Rosenberg and J. Tomeo, 2004, Hedge Fund of Fund Allocations Using a
Convergent and Divergent Strategy Approach, The Journal of Alternative Investments,
Summer.
Gorton, G., F. Hayashi and K. G. Rouwenhorst, 2007, The Fundamentals of Commodity

Futures Returns, Yale ICF Working Paper No. 0708.
Gorton, G. and K. G. Rouwenhorst, 2006, Facts and Fantasies about Commodity

Futures, Financial Analysts Journal, 62(2), March/April.
Ilmanen, A., 2011, Expected Returns: An Investors Guide to Harvesting Market Rewards
(Hoboken, NJ: Wiley).
Spurgin, R., 1999, A Benchmark for Commodity Trading Advisor Performance, Journal
of Alternative Investments, Fall.
305
12
Structural Alpha Strategies
Francisco Blanch; Gustavo Soares and Paul D. Kaplan
Bank of America Merrill Lynch; Macquarie Funding Holding Inc.
and Morningstar, Inc.
Investors can get exposure to market-neutral returns in commodities

commodity alpha in many different ways. Active management is,
of course, one means of obtaining exposure to commodity alpha.
However, as commodities have grown as an asset class, a large
number of rules-based strategies (examples of which will be
provided throughout the chapter) designed to generate market-
neutral returns have emerged. These systematic commodity alpha
strategies exploit structural characteristics of commodity markets,
and will be referred to here as structural commodity alpha strate-
gies. They are not a source of risk-free returns. Their returns are the
reward for taking on risks that other market participants are
unwilling to take on a systematic basis.
The main goal of this chapter is to outline the major investment
themes among these structural commodity alpha strategies and
suggest a simple methodology for combining them into an absolute
return portfolio. After briefly introducing the three major investment
themes in commodities alpha, this chapter will examine each in turn.
First, it will cover curve placement strategies, discussing their ratio-
nale and relationship to storage economics. This section will also
explore how curve placement strategies are generally constructed
and their risks, as well as issues including seasonality and market
segmentation. The next section looks at momentum strategies, going
over the principle behind such strategies and their relationship with
the economic cycle. The section discussed the construction of
momentum strategies using price as a signal, and also using the
307
shape of the forward curve as a signal. It also assesses the distinction

between absolute and relative momentum.
The third section covers volatility strategies, examining the ratio-
nale of such strategies and comparing different implementations
using variance swaps and variance swap calendars and the risk
return profile expected of each implementation. The final main
section discusses how to combine different alpha strategies into a
basket in order to achieve particular riskreturn goals for the overall
portfolio. In the process, it also explores how the approach of
choosing a set of relatively simple, uncorrelated strategies is a means
to mitigate backtesting bias relative to out-of-sample performance.
Finally, we conclude by highlighting the main points considered
throughout the chapter.
In general, there are three major investment themes on structural
commodity alpha: curve placement, momentum and volatility.
Strategies that provide price insurance and liquidity to market
participants on a systematic basis are rewarded with positive
returns. These strategies are not riskless, but can be constructed to be
more or less independent of the factors that affect commodity price.
As a result, their returns have low correlation with market returns.
Even although these strategies are not independent of market funda-
mentals, they cannot be replicated by simply getting exposure to a
broad-based commodity market benchmark. Hence, they constitute
pure commodity alpha.
Curve placement strategies are one of the most basic and popular
ways to generate commodity alpha. In essence, they exploit market
segmentation (ie, the fact that different market participants have
different hedging needs) across different commodity forward
curves, as consumers, producers and index trackers tend to use
different contracts for their hedging needs. Curve placement strate-
gies aim to generate returns by taking advantage of the differences in
hedging needs between producers, consumers and index trackers. In
addition, curve placement strategies are rewarded for providing
liquidity to market participants in less liquid parts of the forward
curve.
Momentum strategies can be used by commodity investors as a

source of alpha in many different ways. Momentum alpha is the
308
STRUCTURAL ALPHA STRATEGIES
reward for taking price risk from market participants ahead of the
market using the fact that commodity prices and inventories follow
persistent fundamental economic trends. In particular, roll returns
are closely linked to inventory cycles. Given that changes in invento-
ries are the differences between supply and demand, momentum in
commodities is ultimately the result of persistence in inventory
levels and changes. As inventories build (or draw) slowly over time,
momentum generates alpha by identifying the commodity markets
that need to create incentives for market participants to balance
markets by moving physical commodities in and out of storage, or
incentivising changes in demand or supply.
Volatility strategies provide insurance to market participants and are

rewarded for taking price risk from market participants that are
unwilling to bear those risks. As in any other derivatives market,
implied volatility in commodity markets serves as the key parameter
for market participants eg, producers, consumers, processors and
investors to match supply and demand for options. However, there
is often a structural imbalance between buyers and sellers of options
in commodity markets. Market participants hedging needs cause
biases in the options markets, offering a source of market-neutral
alpha for investors.
CURVE PLACEMENT STRATEGIES

To understand how alpha can be generated by curve placement
strategies, we need to understand how commodity forward curves
are shaped and how different market participants segment them-
selves across the curves. Moreover, we need to be aware of when and
how these market participants position themselves throughout the
year and the trading flows associated with these positions.
Forward curves and the market value of storage

Commodity forward curves embed not just expectations about
future prices, but also the net costs of carrying physical commodities
over time. Hence, storage and financing costs largely explain the
shape of the curve in most commodity markets. A simple arbitrage
argument shows how the cost of physical storage should be equal to
the difference between forward and spot prices (see Figure 12.1). If
the net cost of physically storing a commodity is higher (lower) than
309
Figure 12.1 Storage economics determines the shape of the forward curve: cost
of financing plus storage costs
102 $/bbl
Risk premium: a producer would accept a lower
100 price than expected in exchange for locking-in his
margins: F0,T S0 = E0(ST) S0 Risk Premium
98
Market expectation of
future prices, E0(ST).
96 Actual future price, F0,T How do we determine them?
94 Actual spot price, S0
92
Purchasing the commodity in the spot market,
90 storing it and selling the future generates
F0,T S0 = cost of storage + financing
Futures expiry
88
22-Feb-11 22-May-11 22-Aug-11 22-Nov-11
the difference between spot and forward prices, owners of the phys-
ical commodity would rather sell (buy) the commodity on the spot
market and buy (sell) it forward than store it. This dynamic would
force spot prices down and forward prices up when the forward
curve is not steep enough to compensate for storage costs. Similarly,
it would force spot prices up and forward prices down if the forward
curve is too steep.
However, storage costs interact with market expectations and the
need of physical players to own the physical commodity. Depending
on market conditions, one of the factors may be more relevant than
the others. For commodities that are hard and expensive to store
such as natural gas, crude oil and lean hogs (a commonly used type
of pork that is traded in Chicago) the cost of storage tends to play
an important role in determining the slope of the forward curve, but
the shape of the forward curve can, at times, deviate widely from the
storage cost implied contango.
But what determines the curvature of the forward curve? That is,
what determines the difference between the 1M2M timespread and
the 3M4M timespread? Using the same physical arbitrage argu-
ment, it should be the cost of financing and storing a commodity for
a month starting in one month versus the cost of financing and
storing a commodity for a month starting three months from now.
310
Hence, if having access to a storage facility now is more valued by

the market than having access to a storage facility only available in
three months, then our cost of storage argument implies a certain
concavity in the shape of the forward curve. Concavity in the shape
of forward curves suggests that the 4M contract should experience
less price decay as time passes than the 2M contract. In other words,
the price of the 4M contract falls less as it becomes the 3M contract
than the price of the 2M falls as it becomes the 1M contract. Curve
placement strategies take advantage of concavity in the forward
curves by rolling long positions in commodity futures further out in
the forward curve versus rolling short positions in commodity
futures closer to maturity. Historically, this strategy has generated
consistent outperformance and a longer date exposure produces a
better risk-adjusted return. These strategies can be implemented
with any set of weights. However, using DJUBS weights has been
the most popular way of implementing the curve placement strate-
gies (see Figure 12.2).
Market segmentation across commodity forward curves

Curve placement strategies exploit differences in the market value of
storage across contract maturities, and can also take advantage of
market segmentation across the forward curve. Producers and
consumers are willing to pay a premium to protect their profit
Figure 12.2 Risk-return of DJ-UBS based curve placement strategies (bubble

sizes and labels are information ratios)
8%
Returns
7%
2.83
6%
5% 3 month forward
2.80
4%
2.53 2 month forward
3%
2% 1 month forward
Volatility
1%
1.0% 1.2% 1.4% 1.6% 1.8% 2.0% 2.2% 2.4%
311
margins against large price movements, as do index hedgers.

However, these market participants tend to favour transactions
across differing tenor segments of the forward curve.
Producers tend to hedge their long positions on the back-end of
the curve, often accepting a lower price than they would expect in
order to secure the profitability of their investments. In contrast,
consumers tend to hedge their short positions on the front-end of the
curve, often accepting to pay a premium for the physical ownership
of the commodity. Finally, index hedgers provide systematic buying
and selling pressure, selling the front-end contract and buying the
second or third most nearby contract, on a recurring basis.
Curve placement strategies benefit from market segmentation
because they roll long positions in contracts further out in the
forward curve (facing the producers as they search to offload their
natural long price risk) and roll short positions in contracts close to
maturity (facing the buying pressure created by consumers and the
index hedgers in the front-end of the forward curve).
Seasonality in curve placement strategies

Seasonality patterns can be found in a variety of commodity markets
such as livestock, refined products, natural gas, grains, sugar
and coffee. Seasonality can then be used to enhance the risk
return profile of curve placement strategies by leveraging the
seasonal patterns of contract liquidity, storage needs and market
segmentation.
Because of the seasonality in production, certain commodities
tend to come into the market at a very specific time of the year. For
example, corn tends to be harvested in the northern hemisphere
where the bulk of the world production is located between the
months of August and November. Inventories reach their lowest
point in Q3, just as the corn harvest starts in the northern hemi-
sphere. Hence, there is seasonality in the market value of storage.
There is also seasonality in the hedging needs of consumers and
producers. For example, corn producers have to decide how much to
invest in terms of seeds, fertilisers and other production inputs
during the planting and growing seasons. In order to fix their
margins, they increase their hedging demand during the planting
season and decrease it during the harvest season. In particular,
producers start closing down their short corn futures positions on
312
the September contract (when the harvest starts coming in) during
the months of June, July and August, exerting some buying pressure
on the September contract during those months.
Risks of curve placement strategies

One of the characteristics of curve placement alpha is that it faces
headwinds when beta commodity investments perform well. It is a
stylised feature of commodity forward curves that spot prices tend to
tilt the forward curve into backwardation as they move higher. As
demand starts outpacing supply, the spot price tends to rise and
inventories tend to draw. Lower levels of inventories should
decrease storage costs, and push down forward prices relative to
spot prices. In fact, a backwardated curve is the markets way of
offering storage holders an incentive to supply the commodity into
the spot market. Therefore, when the market is tightening, spot
prices move up and the forward curve flattens or moves into back-
wardation.
This link between spot prices and the shape of the forward curve
is behind one of the most important characteristics of curve place-
ment alpha: its negative correlation with the market. By being long
contracts with longer maturity and short the contracts on the front
end of the curve, curve placement alpha strategies tend to get hurt
when commodity prices rally. A common solution to this problem is
to use the monthly rebalancing of the curve alpha strategy to
neutralise its exposure to the market by having less notional expo-
sure on the short leg than on the long leg. The result is a
beta-neutralised curve placement strategy. While beta neutralisation
is not a performance enhancement feature, it eliminates the negative
correlation between curve placement alpha and commodity beta. As
a result, investors are left with a purer alpha strategy that is not nega-
tively impacted by commodity price rallies. Investors who are
looking for a more market-neutral strategy should neutralise their
beta exposure in their curve placement alpha strategies.
Another often proposed solution to this problem is the dynamic
allocation across maturities. If curve placement alpha suffers when
forward curves move into backwardation, then it may be better to
dynamically adjust the strategy according to the shape of the
forward curve. For instance, the investor may choose the position in
the forward curve that has the best-implied roll yield (ie, the
313
expected price appreciation/depreciation of the contracts as it

approaches expiry) to the contract with lower maturity. In principle,
this would give us a forward-looking estimate of how much implied
roll costs will be the following month.
However, this seems to be an overly simplistic approach. These
strategies do not seem to work in practice. The fact that these strate-
gies fail to differentiate themselves from static exposure is illustrated
in Figure 12.3. The Dow JonesUBS Roll Select Commodity Index,
which rolls into the futures showing the most backwardation or the
least contango, seems to underperform simple static allocations such
as the Dow JonesUBS 5M Forward Commodity Index.
Perhaps, this is not surprising. The current yield or current
implied roll cost of a contract can only be a good predictor of its
future performance if the shape of the forward curve remains the
same. However, the slope and curvature of commodity forward
curves are not random walks and their best predictors are not their
current values. It is fairly easy to statistically reject the random walk
hypothesis for the slope and the curvature of the forward curve
across most commodities. In fact, the slope and the curvature of
commodity forward curves tend to show a large degree of mean
Figure 12.3 Dynamic curve placement strategies have failed to differentiate

themselves from static exposure
200 2%
Jan-06=100
180 2%
160
1%
140
120 1%
100 0%
80 -1%
60
-1%
40
20 -2%
0 -2%
06 07 08 09 10 11
Monthly outpeformance (rhs) DJUBSF5 DJUBS Roll Select
314
reversion. Hence, yield or implied roll cost strategies are

unlikely to work in practice because they do not even work in theory.
MOMENTUM STRATEGIES
Momentum strategies can be used by commodity investors as a
source of alpha in a range of different ways. For instance, many
managed futures funds (also called commodity trading advisors,
CTAs) employ computer-based algorithms that aim to identify
upward and downward price trends across a variety of markets.
Most of these algorithms work under high frequency and try to take
advantage of statistical patterns and inefficiencies not only in
commodities, but across a many futures markets. Alternatively,
momentum strategies can be profitably implemented in low-
frequency models such as those based on monthly returns.
High-frequency systematic momentum trading (as defined above)
can be a profitable strategy in certain market circumstances, and is
likely to be a diversifying strategy on a broad portfolio of alpha trades.
However, they are hard for investors to access outside of a fund
format. Low-frequency momentum, on the other hand, can be easily
implemented. Most importantly, high-frequency momentum and
low-frequency momentum are not competing strategies, but rather
complementary on a broad basket of commodity alpha strategies.
Where does low-frequency momentum come from? For
commodities such as crude oil, refined products and base metals,
momentum comes from their cyclical nature that is, from the fact
that demand follows the upward and downward trends of the busi-
ness cycle (see Figure 12.4). More broadly, persistence is a direct
consequence of determined demand growth combined with the
inability of production to respond immediately to demand shocks, in
addition to low short-term elasticity of supply. Hence, momentum
results from persistence in the supply and demand fundamentals of
the commodity.
Given that changes in inventories are the differences between
supply and demand, momentum in commodities results from a
sustained trend in the levels of inventories. As such, we should find
an even stronger link to the shape of the forward curve. In fact, statis-
tical analysis shows momentum in the shape of the curve seems to be
more prevalent and easier to detect than momentum on returns (see
Figure 12.5).
315
Figure 12.4 t-statistics of previous month performance for DJ-UBS ER

sub-indices
Aluminium
Copper
Crude oil (WTI)
Heating oil
Nickel
Gasoline
Sugar
DJ-UBS ER
Live cattle
Zinc
Corn
Natural gas
Soybean oil
Soybean Statistically
Cotton
Silver insignificant at the
Wheat 10% level
Lean hogs
Gold
Coffee
-4 -3 -2 -1 0 1 2 3 4
Jul-05 to Jul-10 Jul-00 to Jul-10
Figure 12.5 Degree of persistence on the shape of the forward curve (From
Jan-06 to Dec-10)
Gold 39.2
Silver 24.2
Coffee 15.7
Nickel 15.3
Zinc 13.7
Copper 10.5
Aluminium 9.0
Sugar 9.0 All commodities have
Heating oil 8.9 statistically significant
Cotton 8.4 persistance in the shape
Corn 6.7
Crude oil (WTI) 6.3 of the forward curve
Gasoline 6.0
Natural gas 6.0
Lean hogs 5.6
Soybean 5.5
Soybean oil 5.4
Cocoa 4.9
Live cattle 4.4
Wheat 2.6
0 10 20 30 40 50
Whether using excess returns, momentum or momentum based

on the shape of the curve, momentum strategies exploit persistence
in the levels of inventories. Months of low inventory levels which
are associated with backwardation, low roll costs and rising prices
316
tend to be followed by months of low inventories. Hence, positive

price performance and low roll costs are likely to be persistent.
Similarly, months of high inventory levels which are associated
with contango, high roll costs and falling prices tend to be followed
by months of high inventories. Momentum generates alpha by iden-
tifying the commodity markets that need to create persistent
incentives for the market to balance supplies with demands. The
strategy is then rewarded for taking price risk in anticipation of
future price movements. In that regard, momentum strategies are
not pure alpha strategies designed to be market-neutral at all times,
quite the contrary; in times of price appreciation, momentum strate-
gies would be expected to participate and have notable positive
correlation with the market. In times of price declines, short posi-
tioning would result in negative correlation with the market.
Still, momentum strategies do not have an inherent bias to one
side or another, and in that sense are market-neutral alpha strategies.
The drawback of momentum strategies is that they are only likely to
generate returns when markets are trending, and are less likely to
generate much alpha in range-bound markets.
While price and curve momentum strategies roughly exploit the
same source of alpha, there is an important difference when it comes
to implementation. Momentum can be used as a relative value signal
that generates cross-commodity alpha by selecting which commodi-
ties to go long. For instance, some strategies select the most
backwardated commodities or the least contangoed to construct
a long-only portfolio. These are relative momentum strategies that
allow investors to pick the bestperforming commodities, imposing
the constraint to be 100% long at all times.
However, one can also use momentum as an absolute value indi-
cator ie, as a signal to be used when deciding whether to go long or
short a particular commodity. A portfolio using an absolute
momentum strategy does not need to be 100% long at all times.
Unlike relative momentum strategies, a portfolio using this type of
implementation could be long for some commodities without any
offsetting short positions. Similarly, the portfolio could have only
short positions or stay neutral depending on the combination of
signals it receives.
317
VOLATILITY STRATEGIES
Commodity volatility can also provide a powerful source of alpha
for investors. Just as in any other derivatives market, implied
volatility in commodity markets serves as the key parameter for
market participants such as producers, consumers, processors and
investors to match the supply and demand for options. However,
there is often a structural imbalance between buyers and sellers of
options in most commodity markets.
Market participants hedging needs cause persistent biases in the
commodity options markets, offering a source of market-neutral
alpha for investors. Commercial market participants are natural
buyers of insurance against large price swings ie, buyers of
volatility. Producers and consumers are willing to pay a premium to
protect their profit margins against large price movements. At the
same time, there are few natural sellers of volatility in the
commodity options markets apart from speculators.
Because of the relatively low participation of speculators in
commodity options markets, this imbalance between buyers and
sellers of volatility has helped to create structural alpha opportuni-
ties for investors. For market participants willing to take on price
risk, there is an opportunity to profit from this demand for insur-
ance. Generally, there is high demand for long option positions
among commercial market participants, such as producers,
consumers and distributers.
Option sellers collect the premium of an option and often delta
hedge their exposure to the underlying contract. However, at incep-
tion, option sellers do not know whether the final profits of delta
hedging the position will be positive. The difference between the
option price change and the profit or loss of the underlying delta
hedge is the hedging error that affects the profit and loss (P&L) of
selling options and delta hedging. Hence, option market makers
need to be compensated for the risk of losing money on their delta
hedges.
This hedging risk is a function of the gamma of the option, as
well as the future realised price volatility of the underlying future.
As a result, option prices (and consequently implied volatility) need
to be high enough to compensate market makers for the risk of
volatility realising at high levels over the lifetime of the option. While
the high demand for options from commercial players pushes up
318
implied volatility away from realised volatility, current realised

volatility is an estimate of the risk of delta hedging. Hence, the
implied versus realised volatility spread can be seen a measure of the
implicit supply and demand for volatility in the options market. As
the P&L of delta hedging short options positions is linked to the
spread between implied and subsequently realised volatility, option
market markers then embed a premium on implied volatility over
realised volatility.
Investors can benefit from the imbalances between buyers and
sellers of commodity volatility through variance swaps. The payout
of a variance swap depends directly on the difference between
implied volatility and the subsequently realised volatility of the
underlying asset. Hence, commodity variance swaps can be utilised
to take advantage of the structural premium between implied and
realised volatility in commodities. Systematic variance selling offers
a sources of structural alpha for investors. A variance swap can be
sold short at the close of the day that the previous swap expires so
that short variance positions can be rolled over continually (see
Figure 12.6).
Selling variance is vulnerable to large price movements common
in many commodity markets. One way to mitigate this risk is to
hedge the short variance position with a long variance position with
Figure 12.6 MLCX WTI vol arbitrage excess return index MLCXCVA1 index
5%
215
4%
3% 195
2%
175
1%
0%
155
-1%
-2% 135
-3%
115
-4%
-5% 95
Dec-02 Jun-04 Dec-05 Jun-07 Dec-08 Jun-10 Dec-11 Jun-13
Monthly returns Index levels (rhs)
319
longer maturity, further out in the volatility term structure. Such

calendar spread variance swaps can be used to create consistent
structural commodity alpha strategies in many commodity markets.
Combining strategies with different tenors and then adjusting
notionals to make the exposure less sensitive to changes in implied
vols ie, vega neutral is a classic way of mitigating the tail risk in
short volatility strategies.
Options have convex payouts and their price movements are not
exactly equal to the movements in the underlying delta hedge. The
more convex the option, the harder it is to delta hedge, the higher
risks the option seller faces and the higher the compensation for
bearing those risks should be. Hence, the alpha generated by system-
atically selling variance should be proportional to the degree of
convexity of the option payout function ie, to the options
gamma. As a result, systematically selling variance in short matu-
rity tenors should outperform systematically selling variance in long
maturity tenors. This is so because the short-dated options have
higher gamma than long-dated options, everything else being
constant. Similar gamma strategies can also be used to generate
structural commodity alpha in many commodity markets (see Figure
12.7).
Figure 12.7 MLCXCVSB index with monthly returns (WTI crude oil 1M versus
3M variance swap calendar spreads)
5% 175
4% 165
3%
155
2%
145
1%
135
0%
125
-1%
115
-2%
-3% 105
-4% 95
Dec-02 Jun-04 Dec-05 Jun-07 Dec-08 Jun-10 Dec-11 Jun-13
Monthly returns Index levels (rhs)
320
PUTTING IT ALL TOGETHER

The objective of any alpha portfolio should be to capture the benefits
that a diversified portfolio of different sources of alpha can provide.
In general, a portfolios risk is a function of the number of strategies
held in the portfolio as well as of the correlation between them.
Hence, a portfolio of a few alpha strategies, more or less independent
from each other, generally will produce better risk-adjusted returns
than any single strategy by itself.
A simple example illustrates the power of a diversification in an
alpha portfolio. Suppose we have an alpha strategy (strategy A) that
generates returns of 5% and volatility of 2% per annum, providing an
information ratio of 2.5. At the same time, suppose we have a set of
three uncorrelated strategies (strategies B, C and D), each generating
annual returns of 3% with the same 2% volatility, each of these
providing an information ratio of 1.5. Despite each of those strategies
being far worse than strategy A on an individual basis, it turns out
that an equally weighted portfolio of strategies B, C and D produces
a better allocation in risk-adjusted terms than an allocation to
strategy A.
Backtesting bias
In any historical backtest, it is hard to identify whether the good
historical performance of a strategy is a product of its design or pure
luck. This is a classic problem with backtests and other types of
model-selection algorithm. By searching for the alpha holy grail, we
may end up spuriously choosing a methodology that would have
performed well in that period by sheer luck. Complex strategies may
perform well on a backtested basis not because of any fundamental
reason, but only because their many bells and whistles were chosen
so the strategy would perform well on the backtest in the first place.
In fact, a set of sufficiently complex rules can overfit history and give
any strategy a great backtest.
One way to mitigate the risk of backtesting bias is to always
choose simple implementations of each investment theme. Simple
strategies have fewer degrees of freedom and therefore are likely to
suffer less from the overfitting problem that complex strategies
intrinsically embed. Overfitting rules and parameters are what ulti-
mately generate the backtesting bias. Investors are better off
combining simpler strategies potentially with worse backtested
321
performance, but for which they can understand the sources of

returns than more complex versions of the same strategy.
The three major investment themes in the structural commodity
alpha space outlined above curve placement, momentum and
volatility can be accessed through extremely simple rules-based
strategies. Of course, one could think of many ways to try to improve
the performance of these simple strategies. However, in practice,
many of the more complex strategies only marginally improve the
risk-adjusted returns of the simpler strategies. At the same time,
simpler strategies are less vulnerable to the backtesting bias problem
purely because they have less rules and parameters to be calibrated.
Weaving an alpha basket

The simple example above suggests an easy methodology for
creating high-quality alpha in any asset class. Using simple strate-
gies, investors should create an alpha portfolio that captures
different sources of risk premium. Generally, the diversified alpha
portfolio will produce better risk-adjusted returns than any single
strategy if the strategies are combined in an optimal way.
However, what are the weights that should be given to each alpha
strategy when constructing a portfolio of alpha strategies? The
concept of basket weights is somewhat meaningless for
commodity alpha investments. What really matters is not notional
weight but volatility weight ie, the contribution of each strategy
to the overall risk of the portfolio.
To make alpha strategies comparable, investors can dynamically
adjust the allocation to the strategy in order to target a desired level
of volatility. This technique, called volatility targeting, is a means of
creating a level playing field for different alpha strategies. Because
commodity alpha strategies are typically unfunded, changing the
degree of participation is only limited by the amount of collateral
needed to fund the strategy.
On the level playing field of volatility-targeted strategies, how do
we then best combine different sources of commodity alpha into a
single alpha portfolio? After defining expected returns and a
measure of risk (eg, volatility) for every alpha strategy, we can
construct an efficient frontier by finding the portfolio that maximises
expected returns for a given amount of risk. Because commodity
alpha strategies are unfunded, the weights of a commodity alpha
322
portfolio do not need to add to 100% as they do in the classic

Markowitzs efficient frontier problem.1 In the absence of any weight
constraint, the efficient frontier for commodity alpha portfolios is a
straight line in which all portfolios have exactly the same informa-
tion ratio as that of the maximum-information ratio portfolio (MIP)
of a frontier obtained with portfolio weights constrained to add to
100% (see Figure 12.8).
However, defining expected returns and a measure of risk for
each commodity alpha strategy involves the real problem of trying to
optimise different sources of commodity alpha into a portfolio. That
is, it is a statistical rather than a financial issue. One way to avoid the
statistical difficulties relating to asset allocation decisions is to simply
put statistics aside and follow an appropriate rule of thumb. On the
level playing field of volatility-targeted strategies, an equally
weighted (EW) basket is a straightforward way to combine
commodity alpha strategies. One advantage of the EW basket is its
robustness over time, as the information ratios of the EW basket tend
to be more stable than the ones for individual strategies. However,
the EW basket is more than just a nave diversification technique.
In our selected set of commodity alpha strategies, a case could be
made for all pairwise correlations being zero on average. After all,
Figure 12.8 The application of CAPM to commodity alpha suggests that

investors should only care about the Maximum Information Ratio Portfolio
Efficient frontier: max returns given a targeted level of risk
6.0% Returns
5.5%
MIP: Maximum Information Ratio Portfolio
5.0% with weights adding to 100%
4.5%
4.0%
Portfolios with unconstrained
3.5% weights have the same
information ratio as the MIP
3.0%
Targeted
0.75% 0.95% 1.15% 1.35%
volatility
Weights constrained to 100%
No constraints (weights adding to more than 100%)
No constraints (weights adding to less than 100%)
Source: BofA Merrill Lynch Global Commodity Research
323
Figure 12.9 In practice, the EW basket or the Minimum-Variance Portfolio can

get investors quite close to the efficient frontier
Efficient frontier: max returns given a targeted level of risk
6.0% Returns
5.5%
MVP: Minimum-Variance Portfolio with
5.0% weights adding to 100%
4.5% Efficient frontier

with unconstrained weights
4.0%
3.5% Leveraging up and down

the MVP
3.0%
Targeted
0.8% 1.0% 1.2% 1.4%
volatility
Weights constrained to 100%
No constraints (weights adding to more than 100%)
No constraints (weights adding to less than 100%)
Source: BofA Merrill Lynch Global Commodity Research
the strategies were designed to be more or less independent sources

of alpha. If that is the case, then on the level playing field of volatility-
targeted strategies, we know that the EW basket is one with weights
adding up to 100% with the lowest level of volatility. Formally, if all
pairwise correlations are equal to zero and the volatility of each
strategy is equal to s, then the EW basket is the portfolio with
weights adding to 100% with the lowest level of volatility. Under
those assumptions, this is the minimum-variance portfolio (MVP).
Of course, the weights of a commodity alpha portfolio in the effi-
cient frontier do not need to add to 100%. However, once the MVP is
found, investors can leverage weights up and down proportionally
to achieve any level of targeted volatility on the overall portfolio. In
practice, this technique can get investors quite close to the MIP
without having to ever care about estimating expected returns of
commodity alpha strategies (see Figure 12.9).
CONCLUSION
Systematic commodity alpha strategies attempt to capture different
risk premiums, such as insurance and liquidity, prevalent in
commodity markets. These systematic strategies capture risk
premiums that are structural to commodity markets and therefore
324
likely to be a robust source of market-neutral returns in the long run.

In addition, a portfolio of alpha strategies more or less independent
of each other will yield higher risk-adjusted returns than any stand-
alone, single commodity alpha strategy. This paves the way for a
simple recipe for generating high-quality alpha in commodities.
Investors should pursue simple and easy to understand strategies
that exploit a broad range of sources of alpha. In particular, using
equally weighted baskets of volatility-targeted strategies is a simple
and robust way to construct a long-term allocation to structural
commodity alpha strategies.
Reprinted by permission. Copyright 2013 Merrill Lynch, Pierce,
Fenner & Smith Incorporated. Further reproduction or distribution is
strictly prohibited.
APPENDIX: PUTTING MOMENTUM INTO COMMODITIES

Paul D. Kaplan
Strategies that take a momentum-based long/short approach to

commodity investing serve investors better than long-only strate-
gies. Following weaker investor activity in 2011, investment flows
into commodities grew in 2012, with a net inflow of US$5.3 billion
across all sectors alone in August 2012, with commodity exchange-
traded products (ETPs) one of the fastest-growing asset classes.
Commodity ETPs hit an all-time high of US$207.4 billion total assets
in September 2012. In comparison, commodity ETPs saw inflows of
US$10 billion throughout the entire year of 2011, and the year ended
with total assets in commodities ETPs of US$152 billion.
This surge in assets mirrors equally impressive gains in many
commodities spot prices. Unfortunately, and to the chagrin of many
investors, products linked to commodity indexes often experience
much lower returns. Negative roll yield (which occurs when distant
delivery prices exceed near delivery prices) means that many
investors lose out even as prices rise. In response, a growing number
of commodity investors are eschewing the traditional long-only
approach in favour of alternative strategies that are better able to
manage roll yield.
With the rise of more innovative strategies, there is reason to ques-
tion how well investors are being served by the traditional long-only
commodity indexes either as benchmarks or proxies for investment
325
products. Traditional approaches to representing pure beta expo-

sures work well for stocks and bonds but not so well for the
commodities asset class. In fact, we argue that there is no such
thing as commodity beta. Moreover, we also assert that new passive
strategies that use a momentum-based long/short approach rather
than the long-only approach of the most common commodity
indexes are better benchmarks for active strategies.
NO SUCH THING AS COMMODITY BETA

For many asset classes, it is very easy to take a pure beta exposure.
Multiple asset class proxies are available, many of which are reason-
able substitutes for each other. The Russell 3000, S&P 500 and Dow
Jones Wilshire 5000 indexes, for example, are representative of the
broad stock market and have similar performance characteristics,
just as the Citigroup Broad Investment-Grade (BIG), Barclays Capital
US Aggregate and Merrill Lynch US Domestic Master bond indexes
mirror the wider fixed income market and perform alike. However,
for commodities fewer choices and more disparity exist among the
index options.
NOT ALL INDEXES ARE ALIKE

Figure 12A.1 illustrates the similar risk and return characteristics of
the broad stock and bond indexes and the disparity among the three
traditional commodity indexes the S&P GSCITM Commodity
Index, Dow Jones UBS Commodity Index, and Reuters/Jefferies
CRB Index. When we plot standard deviation and compound annual
return for each index over a common time period (January 1991
September 2012), we see that the nearly identical risk and return
characteristics of both the stock and bond indexes place the plot
points on top of one another. The commodity indexes, however, do
not display the same level of consistency. Dramatic differences in
constituents and weighting schemes as well as rebalancing rules are
likely the cause of the performance differences in the commodities
indexes. The S&P GSCI index, for example, has about double the
weighting to the energy sector as the Dow Jones UBS Commodity
and Reuters/Jefferies CRB indexes and only one third of the
weighting to agriculture.
326
Figure 12A.1 Standard deviation versus compound annual returns for various
indexes
Compound
annual return % Stock indexes
Morningstar long/short commodity
Morningstar Reuters/Jefferies CRB

long-only commodity (inception: 02/01/1994)
Bond indexes Commodity indexes
BarCap US
agg. bond
Dow Jones UBS commodity
Standard
deviation %
Source: Morningstar
BOTH LONG AND SHORT POSITIONS FOR POSITIVE RISK

PREMIUMS
Long-only commodity futures strategies can prove inadequate in
providing investment exposure to commodities, which is why
professional CTAs tend to take both long and short positions in
commodity futures, often based on trends in prices.
SOURCES OF EXCESS RETURN

A futures strategy generates excess return (ie, return in excess of the
risk-free rate) from two sources:
327
1. changes in futures prices; and

2. the roll yield which can be either positive or negative that
results from replacing an expiring contract with a further out
contract in order to avoid physical delivery yet maintain posi-
tions in the futures markets.
A complete understanding of these two sources of return requires an

analysis of three interrelated markets for each commodity:
1. spot market the cash market for the commodity itself;

2. futures market the market for contracts to deliver the
commodity in the future for a price set today; and
3. storage market the market for the service of storing the
commodity on behalf of its owner.
What happens in spot markets is important to futures investors

because changes in spot prices impact futures prices. The storage
market is important because it interacts with the spot market and
influences the slope of the futures price curve, which is the source of
roll yield.
At times of high demand, spot prices will be strong and the
futures price will be lower than the spot price so that the further out
the futures contract, the lower the price. When this is the case, we say
that there is backwardation in the futures market or that the
futures curve is backwardated. Investors who are taking long posi-
tions in futures contracts can realise this compensation monetarily by
replacing the contracts that they are holding with longer-term ones,
thus locking in profits.
This component of excess return realised by investors is referred
to as roll yield. As Figure 12A.2 shows, in backwardated markets
roll yields are positive. Likewise, when the marginal benefits of
owning spot supplies are low, the relationship between time to expi-
ration and the futures price is positive, a condition known as
contango. In contango markets, roll yields are negative because
replacing contracts with ones of later maturity results in locking in a
loss.
When a commodity is scarce spot prices are strong, leading to
backwardation and positive roll yields. Conversely, plentiful spot
supply leads to contango and negative roll yields. Since inventory
328
Figure 12A.2 Futures price curves
Positive
Contract price
roll yield
0 (spot)
months to delivery
Contract price
Negative
roll yield
0 (spot)
months to delivery
conditions in some commodities are slow to adjust due to the time it

takes to increase their production, backwardation or contango could
persist for a period of time, causing investors to consistently experi-
ence positive or negative roll yield over the period. Thus, a passive
investor should benefit from a trend-following strategy that incorpo-
rates roll yield into its signal.
329
ROLL YIELD AND EXCESS RETURN

The effect of roll yield on excess return can be substantial. In fact,
several studies have shown that excess return is attributable
primarily to roll yield, not to changes in futures prices. Long-term
excess returns on commodities that exhibit mean reversion in price
and that tend to trade in contango will generally be negative, and
those that tend to trade in backwardation will generally be positive.
This behaviour can be seen in Figure 12A.3, which shows the rela-
tionship between roll yield and excess returns on the commodities
listed for the 21-year period April 1990September 2011.
Commodities that tended to trade in contango experienced negative
excess return, while those that tended to trade in backwardation saw
positive excess return.
Of particular interest here are natural gas futures. Because the
price of natural gas grew at 4.1% per year over the 21-year period,
one might have expected a natural gas futures index to provide a
comparable rate of return. However, because natural gas futures
traded in contango (and consequently experienced negative roll
yield), the excess return was an abysmal negative 12.5%.
BUILDING A BETTER STRATEGY

Passive strategies that use a momentum-based long/short approach
rather than the long-only approach of the most common commodity
Figure 12A.3 Roll yield and excess return
Positive excess
return
positive roll yield
negative roll yield
Negative excess
return
Gasoline-oil-petroleum
Wheat, hard winter
Soybean meal
Natural gas
Soybean oil
Brent crude
Heating oil
Live cattle
WTI crude
Lean hogs
330
indexes can better serve investors by attempting to capture the full

excess return from a futures strategy. Such passive strategies are also
likely to prove a better benchmark for the active strategies of profes-
sional futures investors.
To make this idea operational, we created a family of commodity
indexes that includes combinations of long commodity futures, short
commodity futures and cash (see Figure 12A.4). The primary index,
called the Morningstar Long/Short Commodity Index, holds
commodity futures both long and short based on momentum
signals. The other indexes are derived from this long/short index.
The family includes a long/flat version, which holds cash in place of
the short positions in the primary version so that investors who do
not want or cannot have short positions can still get some benefits of
a momentum-based long/short strategy. The family also includes a
short/flat version for investors who already have long-only expo-
sure to commodities and want some benefits of the momentum
strategy without having to replicate or drop their long-only
exposure.
We created a set of single commodity indexes to serve as
constituents for the long/short index and the related composite
indexes by calculating a linked price series that incorporates
both price changes and roll yield. The weight of each individual
commodity index in each of the composite indexes is the product
of two factors: magnitude and the direction of the momentum
signal. We initially set the magnitude based on a 12-month
average of the dollar-weighted open interest of the commodity.
We then capped the top magnitude at 10% and redistributed any
overage to the magnitudes for the remaining commodities. The
direction depends partly on the type of composite index and, as
we explain below, partly on the type of commodity in the long/
short index.
In the long/short index each month, if the linked price exceeds its
12-month daily moving average, the index takes a long position in
the subsequent month. Conversely, if the linked price is below its 12-
month moving average, the index takes the short side. An exception
is made for commodities in the energy sector. If the signal for a
commodity in the energy sector is short, the weight of that
commodity is moved into cash that is, we take a flat position.
Energy is unique in that its price is extremely sensitive to geopolitical
331
Figure 12A.4 Morningstar commodity indexes construction

Morningstar commodity indexes construction. The Morningstar commodity index family
consists of five indexes that employ different strategic combinations of long futures, short
futures, and cash. The long/short commodity index is a fully collateralised commodity
futures index that uses the momentum rule to determine if each commodity is held long,
short, or flat
Commodity universe
Morningstar commodity universe
Individual commodity indexes

Short-only
Long/flat
Long/short
Long/flat
Long-only
332
events and not necessarily driven purely by supply/demand

imbalances.
For the remaining indexes, the direction is set as follows:
long-only always long for every commodity;

tions as the long/short index, but
Figure 12A.5 Morningstar commodity indexes: riskreturn profile
s
r d d
r
333
Figure 12A.6 Commodity index correlation index
DOWNSIDE PROTECTION
While all long-only commodity indexes tend to provide strong
protection when the stock market is down and in inflationary envi-
ronments, the Morningstar Long/Short Commodity index limits
downside risk while negotiating ups and downs in the commodity
markets themselves. The Long/Short indexs maximum drawdown
in the February 1991September 2012 period, as seen in Figure 12A.5,
was substantially lower than that of the S&P GSCI and Dow Jones
UBS Commodity indexes. We also compared maximum drawdowns
experienced by the listed indexes during five-year sub-periods
within that overall period, and the Morningstar Long/Short
Commodity index suffered much smaller drawdowns in all sub-
periods. Clearly, a long/short strategy is better equipped to tap into
334
the underlying momentum of commodity prices, thereby limiting

losses in down markets.
THE LONG AND SHORT OF IT

The long-only strategies that dominate the commodity index market
do not best serve investors as investment vehicles or as benchmarks.
Since futures price changes and roll yields are the sources of excess
return, long-only indexes have no way to capture the returns avail-
able from shorting futures when there is downward price pressure
or a positively sloped futures price curve. Long-only indexes
generate negative roll yields when markets are in contango (when
distant delivery prices exceed near delivery prices), and thus can
have negative returns when commodity prices are rising.
Furthermore, since many actively managed CTAs invest in long and
short futures based on momentum trading rules, the long-only
indexes are not appropriate benchmarks, rendering traditional
approaches to representing beta exposure unsuitable.
By using a momentum-based approach that takes into account
both price change and the slope of the futures price curve, these
Morningstar indexes aim to maximise both sources of excess return
price change and roll yield to produce better performance. In addi-
tion, these indexes are logically consistent with the underlying
economics of commodities futures markets, and backtested results
show an attractive risk profile, low downside risk and low correla-
tions to both traditional asset classes and long-only commodity
indexes. As passive investment alternatives, these rules-based
indexes could offer easier access to actively managed commodities
trading strategies.
2013 Morningstar. All Rights Reserved. Used with permission.
Further reproduction or distribution is strictly prohibited.
The views and the opinions expressed here are those of the authors
and do not represent the opinions of their employers. The authors are
not responsible for any use that may be made of the contents of this
chapter. No part of this text is intended to influence investment deci-
sions or promote any product or service.
1 The Markowitz efficient frontier problem is: how do you select a portfolio with the lowest
possible risk given a targeted expected return for the portfolio? The solution was developed
by several authors in the 1950s and 1960s, but Harry Markowitzs 1952 paper Portfolio
Selection (Journal of Finance, 7(1), March) was among the first to address the problem.
335
13
Energy Index Tracking
Kostas Andriosopoulos
ESCP Europe Business School
In this chapter, a geometric average Spot Energy Index (SEI) will be

constructed before its performance is reproduced with stock portfo-
lios. The investment methodology employs two self-adaptive
stochastic optimisation methods, superior to other rival approaches
when applied to this index-tracking problem. To test the perfor-
mance of the tracking baskets, three different rebalancing scenarios
are examined, that also take transaction costs into consideration. It
will be shown that energy can be effectively tracked with stock port-
folios selected by the investment methodology used here.
Passive investment strategies are becoming increasingly popular.
Sharpe (1991) argues that, on average, active managers cannot beat
passive strategies and active trading strategies are a zero-sum game.
Other studies have found that passive strategies outperform active
strategies on average (Malkiel, 1995; Sorenson et al, 1998; Frino and
Gallagher, 2001). In addition, Barber and Odean (2000) claim that in
active trading strategies the presence of high transaction costs, and
sometimes the overconfidence of investors in their predictions,
reduces profits substantially and potentially leads to losses.
One of the most popular forms of passive trading strategies, index
tracking, attempts to replicate/reproduce the performance of an
index. Portfolio managers can choose between two methods. Full
replication, purchasing all the stocks in an index, has some practical
limitations and disadvantages. According to Beasley et al (2003),
replicating exactly an index entails frequent revisions1 to reflect the
updated weightings in the index, leading to high transaction costs.
337
One-to-one replication also suffers from the disadvantage that some

stocks can be very illiquid. For these reasons, many passive strategy
managers prefer the alternative of partial replication, where
managers hold the subset of stocks chosen to replicate the index most
effectively.
Since the early 2000s, impressive gains have been witnessed in
commodity prices. This has attracted investors attention and led to
growth of index investing in the commodity markets. In general,
there are three major ways of investing in a commodity index: first,
by choosing an index and replicating it by following the related rule
book; second, by investing in a fund that replicates the chosen index;
finally, a popular approach is buying the shares of an exchange-
traded fund (ETF) that mimics the commodity index. This trend
toward commodity index investing prompted the first commodity
ETF in November 2004.2 As of January 2010, the market capitalisa-
tion of ETF exceeded US$39 billion. Many other ETFs investing in
physical commodities, futures and commodity-related equities have
followed.
This chapter will propose a new approach that reproduces the
performance of a geometric average SEI by investing only in a subset
of stocks from various equity pools. For the purposes of our analysis,
the Dow Jones Composite Average, the FTSE 100 and Bovespa
Composite indexes, and two pools that include only energy-sector
stocks from the US and the UK, respectively, are used. Daily data are
analysed and the index-tracking problem addressed by two evolu-
tionary algorithms the differential evolution (DE) algorithm and
the genetic algorithm (GA). The performance of the resulting invest-
ment strategy is tested under three different scenarios:
buy-and-hold, quarterly and monthly rebalancing, accounting for
transaction costs.
This chapter has the following structure. The next section will
discuss the importance of commodity index investing and major
innovations, before we present the innovative approach used in this
chapter for replicating the price behaviour of an energy commodity
index using equities. We then explain in more detail the optimisation
model, describing the two evolutionary solution techniques
employed. The following section will outline the methodology used
for developing the constructed energy commodity index, along with
the commodity and equity data used, and discussion of the results
338
ENERGY INDEX TRACKING
from tracking the energy index with the proposed investment

approach will then follow.
COMMODITIES INDEX INVESTING

Commodity indexes have been around for many years, used mostly
for benchmarking and to track spot commodity prices. One of the
first published commodity indexes was the Economists
Commodity-Price index, which started in 1864. Then, in 1957, the
Commodity Research Bureau (CRB) index was established, tracking
spot commodity processes; after undergoing major revisions in its
composition, it is still published today. Nevertheless, since the early
1990s, the development of commodity indexes has witnessed
tremendous changes. The first generation of investable commodity
indexes appeared only in 1991, when the S&P GSCI (originally the
Goldman Sachs Commodity Index) was introduced. In 1998, the
Dow JonesUBS Commodity Index (originally the Dow JonesAIG
Commodity Index) and the Rogers International Commodities Index
(RICI) were both launched. Both the S&P GSCI and the RICI indexes
are heavily weighted towards the energy sector, while the Dow
JonesUBS, because of the rule that no sector can weigh more than
one-third of the index, has energy at its limit; in many instances, this
limit is exceeded between the annual rebalancing periods.
The common characteristic, and a major disadvantage of these
early indexes, is that they invest in commodity futures contracts that
are close to expiration, thus they roll forward their futures positions
more frequently making it very expensive to follow an index-
replication strategy using ETFs. In addition, holding a long futures
position via an index that invests in the front of the curve is subop-
timal, especially latterly, because many commodity futures curves
have been experiencing steep contango (a state when the futures
price curve is upward sloping) at the front end of the curve, which
diminishes the ultimate returns.
This previous observation was the main driver behind the creation
of the so-called second-generation commodity indexes such as the
UBS Bloomberg Constant Maturity Commodity Index and the JP
Morgan Commodity Curve Index. Both of these indexes have a
constant weighting scheme across commodities, but their invest-
ments allocation is spread across several contract expirations within
individual commodities. The DJUBSCI 3-Month Forward index
339
takes a similar approach and invests in contracts farther out the

futures curve, reducing the effect of backwardation (a state when the
futures price curve is downward sloping) or contango as the curve
tends to be flatter for longer maturities. These type of indexes outper-
form the first-generation indexes because, when the front end of the
curve is in steep contango, as has been the case with crude oil, the
losses tend to be mitigated or reversed across the longer maturity
contracts. Nonetheless, the opposite happens when futures markets
are in backwardation, since the concentration usually occurs at the
front end of the curve. It can be argued, however, that the chronology
of the indexes has a significant impact on their construction method-
ology, and hence their performance, as later ones have had the
benefit of improving on the methodology used by previously devel-
oped indexes.
The latest addition to the family of commodities indexes are third-
generation indexes that attempt to improve the returns of the
previous two by incorporating commodities selection, over-
weighting or including only commodities that are expected to
deliver higher returns in the near future, while underweighting or
omitting completely commodities that are expected to perform
poorly. The UBS Bloomberg CMCI Active Index introduced in 2007
and the SummerHaven Dynamic Commodity Index introduced in
2009 are two examples of the third-generation commodity indexes.
The latter index includes 14 equally weighted commodities from a
total of 27, rebalancing its futures portfolio every month using basis
and momentum to identify the greatest possible risk premium. The
former index, on the other hand, uses the discretionary approach of
its research analysts who adjust the component weightings of the
index according. However, these types of indexes carry with them a
new risk since the method of the research analysts used to select the
commodities and their respective weightings can be unsuccessful,
and thus underperform passive indexes.
Commodity indexes attempt to replicate the returns equivalent to
holding long positions in various commodities markets without
having to actively manage the positions. Being uncorrelated with the
returns of traditional assets such as stocks and bonds, commodity
index investments returns provide a significant opportunity to
reduce the risk of traditional investment portfolios. This explains the
economic rationale for including a commodity index investment in
340
institutional portfolios such as those of pension funds and university

endowments. There are now numerous publicly available futures
indexes, with different risk and return profiles, offering exposure to
commodity markets; each of these indexes also offers specific expo-
sure to certain commodity sectors via their traded sub-indexes.
Commodity index investing is still relatively young compared
to other more established asset classes such as stocks and bonds
and we would expect continued interest and innovation by market
players in the coming years.
AN INNOVATIVE APPROACH
The above addresses a question that has received almost no attention
in the literature: can returns of equity portfolios be used to replicate
the performance of physical energy price returns, proxied by a spot
index? The aim of this chapter is to replicate the price behaviour of
direct energy commodity investment using equities. The proposed
approach is based on previous research findings that the returns of
equally weighted long-only portfolios of commodity futures are
similar to those of stocks (Bodie and Rosansky, 1980; Fama and
French, 1987; Gorton and Rouwenhorst, 2006). In addition, after the
2000s, commodities went through a financialisation process,
exposing them to the wider financial shocks (Tang and Xiong, 2010).
The replication method uses two very efficient strategies, the DE
algorithm and the GA, to solve the index-tracking problem for the
constructed SEI. These low tracking-error strategies provide several
advantages to investors: they result in better-diversified portfolios,
make the long-only constraint of a fund manager less binding and, in
general, tend to provide higher returns for equity strategies.
The performance of the SEI is reproduced by investing in a small
basket of stocks picked either from the stocks comprising three well-
known financial indexes, or from two pools of energy-related stocks.
In particular, the cases of the US, UK and Brazilian investors are
considered under the assumption that they want to invest in the SEI
and prefer to access only their local stock markets due to cost savings
and/or better knowledge of the respective markets. They represent
two developed and one developing stock market, with the latter
having its unique energy significance in the global scene. Reforms
and regulations that have taken place in Brazil have brought trans-
parency, sophistication and additional liquidity to its financial
341
markets. Oil and other energy prices influence companies earnings

and thus their stock prices. Hence, based on intuition and previous
research findings, the two pools of energy-related stocks used in the
analysis should perform very well in tracking the SEI. Moreover, the
stocks of various companies operating in other, non energy-related
industries will still be affected by the movements in energy prices.
The methodology implemented can track the SEI or any other bench-
mark index by investing in a basket of stocks that each of the
evolutionary algorithms will determine. Baskets of a maximum of
10, 15 and 20 stocks are selected from the following stock pools: Dow
Jones Composite Average, FTSE 100, Bovespa Composite, and the
two pools of energy-related stocks from the US and the UK stock
markets.
The SEI represents a basket of energy commodities and serves as a
performance benchmark with limited ability for direct investment.
However, the proposed approach provides investors with an option
to track the performance of this SEI using a basket of equities that are
liquid and fully investable. This allows investors to get closer to the
underlying commodity market price trends, something they cannot
achieve using a futures price index. Historically, futures index
returns have lagged price index returns, with this decoupling of
performance being a constant frustration for index investors. For
comparison, the performance of two well-established energy excess
return indexes are reported, namely the Dow JonesUBS Energy
Sub-Index and the Rogers Energy Commodity Index, against the
performance of the constructed SEI and the selected portfolios.
This chapters findings have several positive implications for
investors. They provide a low cost compared to actively managed
funds means of accessing the energy spot markets. In particular,
sector rotation investment managers can benefit from the findings.
By tactically shifting assets, they can over- or under-weigh specific
sectors according to their economic outlook or market objective.
Index tracking and problem formulation

In the search for optimally replicating an index, different studies
(Gaivoronski et al, 2004; Frino and Gallagher, 2001) focus on the
performance deviations of the tracking portfolio ie, the tracking
error. Additionally, single-factor and Markowitz models (Larsen-Jr
and Resnick, 1998; Rohweder, 1998; Wang, 1999) have been used to
342
replicate the performance of an index. Furthermore, the use of the

cointegration concept in building portfolios for index tracking is
highlighted by Alexander and Dimitriu (2002) and Dunis and Ho
(2005).
This chapter follows the approach used in Andriosopoulos et al
(2012) for reproducing the performance of an international market
capitalisation shipping stock index and two physical shipping
indexes by investing only in US stock portfolios. First, the tracking
error is measured through the root mean square error (RMSE) crite-
rion. In particular, is assumed that there exist price data on N stocks
and the price of an index over an (in-sample) time period [1, 2, , T].
The goal is to create a tracking portfolio consisting of at most K stocks
(K < N) that replicates, as closely as possible, the index for an (out-of-
sample) period [T + 1, T + t]. The replication error of the tracking
portfolio is defined as follows:
T
2
RMSE = ! (r ! R )
t t /T (13.1)
t=1
where rt and Rt are the returns for the tracking portfolio and the
index, respectively.
Second, except for the replication error, the return of the tracking
portfolio is also of interest. To this end, the mean excess return (ER) is
considered over the benchmark index, defined as follows:
T
ER = " ( rt ! Rt ) /T (13.2)
t=1
Let Pit denote the price of stock i at time t, C the available capital and
xi the number of units bought of stock i. The complete formulation of
the objectives and constraints used to solve the index tracking
problem can then be expressed as follows:
Minimize: f = ! ! RMSE " (1! ! ) ! ER (13.3)

Subject to:
N
!P iT xi = C (13.4)
i=1
zi! C ! PiT xi ! ziC "i = 1,..., N (13.5)
343
!z i !K (13.6)
i=1
xi ! 0, zi ! {0,1} !i = 1,..., N (13.7)
where 0 l 1 is a user-defined parameter that outlines the trade-

off between the two objectives (tracking error and excess return). In
the case l = 1, the tracking portfolio has as its objective to minimise
the tracking error (pure index tracking), whereas when l = 0, the
portfolios goal is to maximize the excess return. Constraint 13.4
guarantees that the value of the portfolio at the end of the in-sample
period is equal to the available capital C. This budgetary limitation
ensures that for all alternative tracking portfolios an identical
amount C is invested at the beginning of the out-of-sample period.
Constraint 13.5 associates a binary variable zi to each stock i, which
is used to consider whether stock i is included in the tracking port-
folio (zi = 1) or not (zi = 0). The parameter e is used to impose a
lower bound on the proportion of the capital invested in each stock
(in this study e is equal to 0.01). Finally, constraint 13.6 defines the
maximum number of stocks K that can be included in the tracking
portfolio.
Evolutionary solution techniques

The optimisation model of equations 13.313.7 is a complex combi-
natorial problem that is difficult to solve with analytical techniques.
Thus, evolutionary algorithms have become particularly popular in
this context. Evolutionary algorithms were first used for addressing
the index-tracking problem by Goldberg (1989), who apply a genetic
algorithm for index replication. More recent applications of genetic
algorithms in index-tracking and portfolio optimisation can be found
in the works of Oh et al (2005), Chang et al (2009) and Soleimani et al
(2009). Beasley et al (2003) propose an evolutionary population
heuristic, accounting for transaction costs and the possibility for revi-
sion of the tracking portfolio. Their results indicate that deriving the
optimal portfolio directly from past data and not from the distribu-
tion of stock returns ultimately achieve better results. Maringer and
Oyewumi (2007) apply DE for tracking the Dow Jones Industrial
Average assuming different cardinality constraints in their selected
344
portfolios. They report that the maximum number of stocks included

in the tracking portfolio must be roughly 50% of the benchmark
index to achieve good results; any additional stocks only marginally
improve the algorithms performance. The DE algorithm has also
been used in other studies using hybrid and multi-objective schemes
(Krink et al, 2009; Krink and Paterlini, 2011), as well as in the context
of loss aversion (Maringer, 2008) and mutual fund replication
(Zhang and Maringer, 2010). Other proposed algorithmic proce-
dures include immune systems (Li et al, 2011), hybrid algorithms
(Ruiz-Torrubiano and Surez, 2009; Scozzari et al, 2012), robust opti-
misation (Chen and Kwon, 2012) and mixed-integer programming
formulations (Canakgoz and Beasley, 2008; Stoyan and Kwon, 2010).
An overview of different methods can be found in Woodside-
Oriakhi et al (2011).
In the context of this chapter, the DE algorithm and a genetic algo-
rithm are employed. Both are well established in the computational
intelligence literature, easy to implement and well suited for
complex financial optimisation problems, particularly in the context
of index tracking and constrained portfolio optimisation. The appli-
cation of both algorithms enables the examination of the robustness
of the results under different solution approaches.
GAs are probably the most popular evolutionary techniques. They
are computational procedures that mimic the process of natural
evolution for solving complex optimisation problems (Goldberg,
1989). A GA implements stochastic search schemes to evolve an
initial population (set) of solutions through selection, mutation and
crossover operators until a good solution is reached.
Similarly to the GA framework, DE is also a stochastic optimisa-
tion method. It was developed by Storn and Price (1995) as an
alternative to existing metaheurtistic approaches, and it is well
suited to continuous optimisation problems. According to Storn and
Price (1997), compared to other rival approaches, the main advan-
tages of DE include its fast convergence, the use of a small set of
tuning parameters, its reduced sensitivity to the initial solution
conditions and its robustness. Overall, comparisons on various
benchmark problems show that DE is superior when compared to
other evolutionary algorithms (Sarker et al, 2002; Sarker and Abbass,
2004).
Both algorithms are implemented with a real-valued solution
345
representation scheme. In particular, each solution is represented by

a real-valued vector x N, where N is the number of stocks in the
sample. The largest positive elements of x are used to identify the
stocks comprising the tracking portfolio,3 and after normalisation (to
sum up to 1) they define the corresponding stock weights (w1, ,
wN). The number of units bought from each stock can then be speci-
fied as xi = Cwi / PiT. The appendix to this chapter provides a brief
description of the implementations of the two evolutionary methods
used here. The parameters of the algorithms were calibrated after
experimentation in order to achieve a good balance between the
quality of the results and the solution times. The selected parameters
are summarised in Table 13A.1.
BENCHMARK ENERGY INDEX, SPOT AND EQUITY DATA

Because many commodities lack centralised trading, the most reli-
able spot prices are for those that trade active and liquid futures
contracts, since these are typically used as a pricing benchmark. In
the case of energy commodities, the Nymex is the worlds largest
futures exchange. Initially, a spot price energy index is constructed,
constituting daily prices of the following six energy commodities,
which also trade futures contracts on the Nymex:
1. Heating Oil, New York Harbour No. 2 Fuel Oil, quoted in US

dollar cents/gallon (C/gal);
2. Crude Oil, West Texas Intermediate (WTI) Spot Cushing,
quoted in US dollars/barrel;
3. Gasoline, New York Harbour Reformulated Blendstock for
Oxygen Blending (RBOB), quoted in US C/gal;
4. Natural Gas, Henry Hub, quoted in US dollars/million British
thermal units (Btus);
5. Propane, Mont Belvieu Texas, quoted in US C/gal; and
6. PJM, Interconnection Electricity Firm on Peak Price Index,
quoted in US dollars/megawatt hour.
The SEI is constructed as an unweighted geometric average of the

individual commodity ratios of current prices to the base period
prices, set at January 31, 2006, until February 1, 2010. The base date
for the SEI is the same date that the equity sample is obtained.
Considering that the boom in commodity index investing is a rela-
346
tively new phenomenon, more recent data are utilised to test the
proposed investment strategy. The indexs construction method-
ology is similar to that of the world-renowned CRB Spot Commodity
Index. The SEI is designed to offer a timely and accurate representa-
tion of a long-only investment in energy commodities using a
transparent and disciplined calculation.
Geometric averaging provides a broad-based exposure to the six
energy commodities, since no single commodity dominates the
index. It also helps increase the index diversification by giving even
the smallest commodity within the basket a reasonably significant
weight. Gordon (2006) finds that a geometrically weighted index is
preferred to alternative weighting schemes, because the daily rebal-
ancing allows the index not to become over- or underweighted. This
avoids the risks that other types of indexes are subject to, such as
potential errors in data sources for production, consumption,
liquidity or other errors that could affect the component weights of
the index. Furthermore, through geometric averaging the SEI is
continuously rebalanced, which means that the index constantly
decreases (increases) its exposure to the commodity markets that
gain (decline) in value, thus avoiding the domination of extreme
price movements of individual commodities. As Erb and Harvey
(2006) point out, the indexes that rebalance annually eventually
become trend followers because commodity prices movements
constantly change the weightings, whereas those that rebalance daily
stay closer to the original intent of the index. In addition, Nathan
(2004) shows that the indexes that use geometric rebalancing, and
thus rebalance their weightings daily, generally exhibit lower
volatility.
The mathematical specification used to calculate the geometric
average SEI is the following:
1
" i Pi %n P1 P 2 Pi
SEIt = $! ti ' ! 100 = n t1 ! t2 !! ti ! 100;i = 1, 2,,6;n = 6 (13.8)
i=1 P P0 P0 P0
# 0 &
where, SEIi is the index for any given day, i represents each one of the
six commodities comprising the index, Pti is the price of each
commodity for any given day, and P0i is the price of each commodity
in the base period.
The SEI provides a stable benchmark structure for the index,
making SEI suitable for institutional investment strategies. The
347
stable composition of the index is an important element, because

when the composition of an index changes over time, the average
return of the index does not equal the return of the average index
constituent, especially when indexes are equally weighted. The latter
makes historical index performance a bad proxy to prospective index
returns, thus distorting the information that investors seek (Erb and
Harvey, 2006). Moreover, it is a better means for evaluating the
movement in energy commodity prices because it is based on spot
prices and not on prices for future delivery that are subject to roll
yields driven by contango and backwardation. The equity data
includes daily prices for stocks picked from the Dow Jones
Composite Average, FTSE 100 and Bovespa Composite indexes. The
equity dataset also includes stocks from a unique pool of energy-
related stocks from the US and UK stock markets. The selection of the
equities included in the two pools is made according to the Industry
Classification Benchmark (ICB) jointly developed by Dow Jones and
the FTSE (see Appendix at the end of this chapter). In the sample
used, the two filtered pools include all stocks from the US and UK
stock markets that are engaged in the various phases of energy
production and processing, listed in the following four sectors: oil
and gas producers; oil equipment, services and distribution; alterna-
tive energy; and electricity. After applying the filtering procedure to
the US and UK stock markets, two energy-related stock pools are
constructed, hereafter named US Filter and UK Filter, respectively.
To test the proposed heuristic approach and the efficiency of both
the DE and GA as index-tracking methodologies, five datasets are
selected. All stock prices are closing prices adjusted dividends
according to the annualised dividend yield, and they are all obtained
on a daily basis for the period January 31, 2006 to February 1, 2010
from Thomson Financial Datastream. All stock prices are in US
dollars, thus reflecting the local currency exchange rate against the
US$ at every point in time for the period examined. Should a
company cease trading due to an event (merger, bankruptcy, etc)
within the test period, it is dropped from the sample that is why the
total number of stocks in the FTSE 100 and Bovespa pools is less than
the total number of stocks included in each index. Moreover, after
adjusting for all US and UK bank holidays, 1,008 observations are
sorted to calculate daily returns for each stock. Considering 252
trading days in a calendar year, the heuristic approach is tested
348
under various assumptions by selecting the first year as the in-

sample period and the last three years as the out-of-sample period.
The final five datasets have the following number of stocks: N = 41
(UK Filter), N = 53 (Bovespa Composite), N = 65 (Dow Jones
Composite Average), N = 77 (US Filter) and N = 97 (FTSE 100 index).
TRACKING THE SPOT ENERGY INDEX

The performance characteristics of the proposed strategy are exam-
ined. The stocks picked by both the DE and the GA are used to track
the performance of the SEI. The initial capital of the investment port-
folio is set equal to C = US$100,000, where both the DE and the GA
converge at the end of the in-sample period. In the empirical
analysis, tracking portfolios consisting of maximum K stocks are
used with K = 10, 15 and 20. Three different trade-offs between
tracking error and excess return are also considered, with l = 0.6, 0.8
and 1, thus moving from maximising excess return to minimising
tracking error. The heuristic is then repeated 10 times with the same
set of parameters per run, from which the best solution is chosen.
Figure 13.1 displays the SEI against quarterly rebalanced portfo-
lios selected from the DE and GA, respectively. The portfolios consist
of a maximum of 15 stocks, the FTSE 100, DJIA, Bovespa and UK
Filter and US Filter, respectively; results are shown for l = 1. Looking
at the figures, it is clear that during and towards the end of the reces-
sion period, the benchmark index can be best tracked with the
Bovespa baskets, followed by the UK Filter baskets; whereas, during
the last year (2010) it is the US Filter and DJIA baskets that perform
better. The portfolios comprised of optimally selected energy-related
stocks can successfully track the SEI, generating similar returns for
most of the out-of-sample period. The US Filter and UK Filter results
verify that, when energy-related stocks are selected, they can better
replicate the risk and return trade-off of the SEI. The same applies for
the Bovespa baskets, since the Brazilian stock exchange has a large
number of energy- and commodity-related listed companies that
would closely follow any developments in the international energy
markets. In addition, between the DE and GA selected portfolios,
from the graphs it seems that the latter ones can follow more closely
the performance of the SEI, achieving highest excess returns for the
final out-of-sample year.
Table 13.1 presents the RMSEs and the mean excess returns of
349
Figure 13.1 Out-of-sample tracking of the SEI with the Bovespa, DJIA, FTSE 100,
UK Filter and US Filter baskets respectively; = 0.8, with maximum 15 stocks in
the basket, rebalanced quarterly
100K Portfolios Q_reb K15L08 (DEA)
200000
180000
160000
140000
120000
100000
80000
60000
40000
20000
0
Feb-07
Apr-07
Jun-07
Aug-07
Oct-07
Dec-07
Feb-08
Apr-08
Jun-08
Aug-08
Oct-08
Dec-08
Feb-09
Apr-09
Jun-09
Aug-09
Oct-09
Dec-09
Feb-10
100K Portfolios Q_reb K15L08 (GA)
200000
180000
160000
140000
120000
100000
80000
60000
40000
20000
0
Feb-07
Apr-07
Jun-07
Aug-07
Oct-07
Dec-07
Feb-08
Apr-08
Jun-08
Aug-08
Oct-08
Dec-08
Feb-09
Apr-09
Jun-09
Aug-09
Oct-09
Dec-09
Feb-10
Bovespa DJIA FTSE UK FILTER US FILTER SEI
both the genetic and differential evolution algorithms employed,

under all three rebalancing strategies: buy-and-hold, monthly and
quarterly rebalancing. Using formal statistical evaluation criteria, the
better tracking performance of the UK Filter and US Filter baskets is
also confirmed. In terms of the competing portfolios RMSEs, the DE
is more consistent across the various portfolios, whereas the GA
selects portfolios that exhibit larger differences between the worst-
and best-performing ones. Additionally, in general, GA tends to
select portfolios that have fewer tracking errors and thus track better
350
the benchmark index when compared to the ones selected from the
DE.
Another interesting observation is that, although the RMSEs are
improved when rebalancing occurs, increasing the frequency from
quarterly to monthly has only a marginal effect. These results are
more profound for the portfolios selected by the DE, and align with
Dunis and Ho (2005), who find that, when comparing alternative
rebalancing frequencies, a quarterly portfolio update is preferable to
monthly, semi-annual or annual reallocations. In terms of their
excess returns, in most cases the portfolios selected by the GA tend to
outperform the ones selected by the DE. The UK Filter and US Filter
baskets, which also have the lowest tracking errors (see panels D and
E on Table 13.1), have excess returns that in some cases are positive,
indicating that the selected portfolios, on average over the out-of-
sample period, outperform the SEI.
In the case of the US Filter baskets selected by the GA, the index is
constantly outperformed in terms of excess returns (8.10% for K = 20
and l = 0.6 under monthly rebalancing, and 6.14% for K = 15 and l =
0.6 under quarterly rebalancing); there is only one exception for both
rebalancing frequencies, in which l = 1 and K = 10 when the portfo-
lios underperform the index. This is an indication that the trade-off
criterion does work, and leads to portfolios that compromise any
excess return over a better tracking performance as expressed by the
smaller RMSEs. Thus, taking into account the fact that commodity
indexes performed better compared to the financial indexes over the
three-year out-of-sample period (except the Bovespa Composite),
with the methodology employed the performance of the SEI is
closely replicated, and in the case of the energy-related stock portfo-
lios, the benchmark index is even outperformed.
For Table 13.1, panels A, B, C, D and E report the out-of-sample
daily RMSE and mean daily percentage excess returns, as defined in
equations 13.9 and 13.10, respectively. Under both rebalancing
strategies, the weights of the tracking portfolios are estimated based
on the available data in the rolling window in-sample period (one
year) every month and quarter, respectively. Portfolios returns are
adjusted for transaction costs of 0.5% for each transaction.
In terms of the riskreturn trade-off (l), it is observed that results
are very similar among portfolios where l = 0.8 and 1. In most cases,
the riskreturn trade-off criterion tends to perform well, selecting
351
352

Table 13.1 Out-of-sample index tracking performance of the selected portfolios.
No rebalance Monthly rebalance Quarterly rebalance
RMSE Mean ER (%) RMSE Mean ER (%) RMSE Mean ER (%)
(K) () DE GA DE GA DE GA DE GA DE GA DE GA
Panel A: Bovespa
10 0.6 0.0346 0.0344 0.0136 0.0324 0.0331 0.0329 0.0432 0.0104 0.0333 0.0332 0.0389 0.0134
0.8 0.0343 0.0359 0.0176 0.0347 0.0330 0.0326 0.0480 0.0471 0.0332 0.0329 0.0438 0.0416
1 0.0343 0.0362 0.0189 0.0133 0.0330 0.0327 0.0545 0.0689 0.0333 0.0332 0.0472 0.0236
15 0.6 0.0345 0.0359 0.0161 0.0239 0.0331 0.0327 0.0427 0.0063 0.0333 0.0332 0.0411 0.0148
0.8 0.0343 0.0361 0.0181 0.0334 0.0330 0.0327 0.0487 0.0298 0.0332 0.0331 0.0431 0.0280
1 0.0343 0.0356 0.0180 0.0238 0.0330 0.0327 0.0533 0.0418 0.0332 0.0333 0.0442 0.0312
20 0.6 0.0345 0.0354 0.0148 0.0233 0.0331 0.0331 0.0436 0.0094 0.0333 0.0335 0.0417 0.0209
0.8 0.0343 0.0358 0.0186 0.0329 0.0330 0.0327 0.0488 0.0052 0.0332 0.0333 0.0427 0.0000
1 0.0343 0.0357 0.0164 0.0284 0.0330 0.0328 0.0541 0.0346 0.0333 0.0334 0.0461 0.0210
Panel B: DJIA
10 0.6 0.0319 0.0328 0.0232 0.0257 0.0318 0.0315 0.0479 0.0115 0.0319 0.0319 0.0302 0.0243
0.8 0.0319 0.0330 0.0238 0.0210 0.0318 0.0316 0.0511 0.0312 0.0318 0.0318 0.0323 0.0273
1 0.0319 0.0330 0.0249 0.0218 0.0318 0.0313 0.0522 0.0274 0.0319 0.0317 0.0314 0.0172
15 0.6 0.0320 0.0329 0.0244 0.0200 0.0319 0.0315 0.0503 0.0332 0.0319 0.0318 0.0297 0.0172
0.8 0.0319 0.0330 0.0240 0.0250 0.0318 0.0314 0.0515 0.0244 0.0319 0.0319 0.0311 0.0192
1 0.0319 0.0328 0.0246 0.0239 0.0318 0.0313 0.0515 0.0410 0.0319 0.0319 0.0314 0.0283
20 0.6 0.0319 0.0328 0.0228 0.0251 0.0319 0.0315 0.0514 0.0239 0.0319 0.0319 0.0313 0.0005
0.8 0.0319 0.0329 0.0235 0.0289 0.0318 0.0315 0.0529 0.0300 0.0319 0.0318 0.0301 0.0332
1 0.0319 0.0328 0.0253 0.0323 0.0318 0.0313 0.0505 0.0344 0.0319 0.0317 0.0308 0.0051
Panel C: FTSE 100
10 0.6 0.0315 0.0318 0.0450 0.0359 0.0309 0.0299 0.0597 0.0260 0.0308 0.0303 0.0438 0.0106
0.8 0.0317 0.0316 0.0469 0.0246 0.0309 0.0302 0.0701 0.0416 0.0309 0.0305 0.0475 0.0255
1 0.0316 0.0314 0.0495 0.0193 0.0310 0.0300 0.0735 0.0635 0.0310 0.0307 0.0461 0.0334
15 0.6 0.0315 0.0318 0.0512 0.0253 0.0309 0.0303 0.0674 0.0327 0.0308 0.0303 0.0468 0.0180
0.8 0.0316 0.0313 0.0477 0.0220 0.0309 0.0302 0.0634 0.0449 0.0309 0.0306 0.0416 0.0127
1 0.0316 0.0312 0.0490 0.0175 0.0310 0.0303 0.0699 0.0682 0.0310 0.0306 0.0456 0.0349
20 0.6 0.0315 0.0317 0.0507 0.0271 0.0309 0.0303 0.0705 0.0311 0.0308 0.0305 0.0442 0.0092
0.8 0.0316 0.0313 0.0484 0.0297 0.0310 0.0303 0.0681 0.0656 0.0309 0.0305 0.0445 0.0145
1 0.0316 0.0313 0.0492 0.0245 0.0310 0.0301 0.0679 0.0600 0.0310 0.0306 0.0449 0.0208
Panel D: UK Filter
10 0.6 0.0318 0.0309 0.0900 0.0834 0.0299 0.0294 0.0712 0.0019 0.0300 0.0296 0.0681 0.0032
0.8 0.0315 0.0312 0.0818 0.0834 0.0300 0.0290 0.0680 0.0725 0.0301 0.0296 0.0611 0.0412
1 0.0317 0.0307 0.0809 0.0751 0.0300 0.0292 0.0713 0.1371 0.0301 0.0297 0.0632 0.1049
15 0.6 0.0312 0.0309 0.0825 0.0519 0.0299 0.0294 0.0782 0.0427 0.0300 0.0298 0.0711 0.0341
0.8 0.0313 0.0309 0.0847 0.0408 0.0300 0.0293 0.0720 0.0501 0.0300 0.0296 0.0707 0.0410
1 0.0313 0.0308 0.0846 0.0531 0.0300 0.0293 0.0782 0.1083 0.0301 0.0297 0.0601 0.0459
20 0.6 0.0311 0.0305 0.0796 0.0586 0.0299 0.0297 0.0764 0.0508 0.0300 0.0299 0.0717 0.0446
0.8 0.0311 0.0303 0.0858 0.0451 0.0299 0.0294 0.0752 0.0790 0.0300 0.0298 0.0697 0.0391
1 0.0311 0.0304 0.0763 0.0516 0.0300 0.0295 0.0747 0.0794 0.0301 0.0296 0.0676 0.0494
Panel E: US Filter
10 0.6 0.0307 0.0329 0.0258 0.0442 0.0306 0.0297 0.0449 0.0710 0.0309 0.0307 0.0364 0.0249
0.8 0.0308 0.0321 0.0265 0.0780 0.0309 0.0295 0.0603 0.0607 0.0310 0.0300 0.0345 0.0240
1 0.0309 0.0318 0.0234 0.0314 0.0310 0.0294 0.0688 0.0278 0.0310 0.0298 0.0367 0.0172

15 0.6 0.0307 0.0321 0.0246 0.0581 0.0309 0.0306 0.0497 0.1241 0.0310 0.0308 0.0322 0.0614
0.8 0.0308 0.0327 0.0244 0.0511 0.0309 0.0296 0.0575 0.0212 0.0310 0.0301 0.0336 0.0016
1 0.0308 0.0322 0.0254 0.0566 0.0309 0.0295 0.0648 0.0027 0.0310 0.0302 0.0342 0.0204
20 0.6 0.0307 0.0327 0.0261 0.0668 0.0309 0.0301 0.0510 0.0810 0.0310 0.0308 0.0274 0.0345
0.8 0.0308 0.0319 0.0251 0.0320 0.0309 0.0296 0.0603 0.0210 0.0310 0.0303 0.0329 0.0369
1 0.0307 0.0311 0.0226 0.0649 0.0309 0.0294 0.0662 0.0071 0.0310 0.0301 0.0352 0.0126
353
portfolios with higher returns and also relatively higher RMSEs.

Moreover, the portfolios selected by the GA tend to be more consis-
tent when the riskreturn trade-off rule is applied, compared to the
ones selected by the DE. Overall, when considering both the tracking
performance and the excess returns of the various portfolios, those
with l = 0.8 should be preferred. As far as the criterion regarding the
maximum number of stocks is concerned, in all three rebalancing
scenarios, portfolios with K = 10 tend to perform worst in terms of
RMSEs, but do slightly better in terms of excess returns, for both the
DE and GA selected portfolios. This is also an indication that the
more stocks that are included in the portfolio, the higher the transac-
tion costs when a rebalancing occurs. Overall, it is suggested that
portfolios with a maximum of 15 stocks should be selected, as there
still seems to be a valuable compensation for the additional informa-
tion and diversification when rebalancing, against the extra
rebalancing costs.
According to the results, for both algorithms, monthly rebalancing
is overall the best option in terms of RMSEs, closely followed by
quarterly rebalancing, whereas when looking at excess returns, quar-
terly rebalancing appears to improve portfolio performance. The
return of a buy-and-hold portfolio may be higher than that of a rebal-
anced portfolio when transaction costs are considered, but it is
important to determine the source of the higher return whether it is
greater capital efficiency as expressed by a higher Sharpe or informa-
tion ratio, or greater risk. Plaxco and Arnott (2002) showed that
rebalanced portfolios typically have higher Sharpe ratios than buy-
and-hold portfolios, a finding that suggests that the possible
outperformance of a buy-and-hold portfolio may be the result of
greater risk. Results are more apparent for the GA portfolios, as for
the DE portfolios the difference between monthly and quarterly
rebalancing is only marginal. In the case of the UK Filter basket
picked by the GA, there is an obvious difference in performance
when rebalancing quarterly as opposed to monthly rebalancing. A
more in-depth analysis comparing the portfolios information ratios
is presented in the following section. On average, based on the
results from Table 13.1, K = 15 and l = 0.8 is the most desirable
combination providing the best results for most tracking portfolios.
It is, of course, up to the investors riskreturn appetite to decide
whether rebalancing the portfolio quarterly, which comes with an
354
extra cost, is better than no rebalancing at all. The same applies

regarding whether l = 0.8 should be used instead of the more risky
trade-off when l = 0.6.
Statistical properties of selected portfolios

Table 13.2 presents some distributional statistics of the selected port-
folios returns under the quarterly rebalancing scenario.4 Also, in
panel F, the statistics and relevant performance measures for the
following indexes are reported for comparison reasons: two total
return energy commodity indexes the DJ UBS-Energy and Rogers
Energy Commodity; the three stock indexes from which stocks were
drawn to construct the tracking portfolios Bovespa, DJIA and FTSE
100; and, finally, the most commonly used benchmark in the finance
industry, the S&P 500. According to the historical annualised volatil-
ities for the out-of-sample period, the SEI is more volatile than the DJ
UBS-Energy and Rogers Energy Commodity Indexes 48.40% as
compared to 36.21% and 41.11% respectively. The respective
volatility of the equity indexes is in the range of 2738%. However,
when comparing the information ratios, only the Bovespa index is
able to generate a better riskreturn performance compared to the
SEI.
Table 13.2 presents the annualised returns and volatilities of the
tracking portfolios, the skewness and kurtosis, the correlation coeffi-
cient between the returns of the benchmark index and the portfolio
that is used each time to replicate this benchmark, and the informa-
tion ratio under the no rebalancing strategy. Panels A, B, C, D and E
represent the portfolios that include stocks picked each time from the
Dow, FTSE 100, Bovespa, UK Filter and US Filter stock pools. Panel F
presents, for comparison, the relevant performance measures for two
total return energy commodity indexes, the DJ UBS-Energy and
Rogers Energy Commodity; for the three stock indexes from which
stocks were drawn to construct the tracking portfolios, Bovespa,
DJIA and FTSE 100; and the S&P 500.
Moving from no rebalancing to monthly rebalancing, the informa-
tion ratios tend to go down in all cases, except in the case of the US
Filter baskets for GA, and that of the UK Filter baskets for both DE
and GA. This can be explained by the higher transaction costs, which
have a greater impact on the portfolios returns, especially in falling
markets. It can be argued that when rebalancing, the additional
355
356

Table 13.2 Distributional statistics of portfolios daily returns under quarterly rebalancing
An. Ret (%) An. Vol. (%) Skewness Ex. Kurtosis Correl. Info Ratio
(K) () DE GA DE GA DE GA DE GA DE GA DE GA
Panel A: Bovespa
10 0.6 6.79 6.38 35.68 38.32 0.572 0.588 7.688 7.146 23.76 27.67 0.185 0.064
0.8 8.04 7.48 35.39 36.15 0.541 0.499 7.696 7.198 23.72 26.04 0.209 0.200
1 8.88 2.94 35.49 37.28 0.537 0.565 7.846 7.791 23.62 26.46 0.225 0.113
15 0.6 7.36 0.73 35.72 38.38 0.578 0.516 7.699 7.113 23.84 28.06 0.196 0.071
0.8 7.86 4.05 35.49 37.33 0.548 0.620 7.910 7.932 23.79 26.89 0.206 0.134
1 8.14 4.87 35.45 36.76 0.532 0.461 7.734 7.889 23.65 25.36 0.211 0.149
20 0.6 7.49 8.27 35.73 38.45 0.570 0.494 7.661 7.896 23.95 26.36 0.199 0.099
0.8 7.77 3.01 35.42 37.53 0.544 0.481 7.675 7.498 23.57 26.21 0.204 0.000
1 8.62 2.29 35.50 37.69 0.534 0.485 7.801 8.467 23.64 25.94 0.220 0.100
Panel B: DJIA
10 0.6 4.61 3.13 19.76 22.72 0.543 0.329 12.944 9.405 8.96 13.36 0.151 0.121
0.8 5.14 3.87 19.79 22.40 0.563 0.444 13.201 9.707 9.13 13.44 0.161 0.136
1 4.90 1.33 19.76 22.87 0.630 0.437 13.884 10.343 8.97 14.63 0.156 0.086
15 0.6 4.48 1.33 19.85 22.44 0.536 0.405 12.659 10.195 9.01 13.63 0.148 0.086
0.8 4.83 1.83 19.80 23.63 0.563 0.210 13.169 8.742 9.04 14.64 0.155 0.095
1 4.91 4.12 19.87 24.36 0.600 0.475 13.712 12.793 8.97 15.65 0.156 0.141
20 0.6 4.87 2.88 19.84 22.41 0.543 0.335 12.801 7.553 9.00 12.49 0.156 0.002
0.8 4.58 5.36 19.83 24.40 0.542 0.355 13.054 9.969 9.07 16.10 0.150 0.165
1 4.75 1.72 19.86 23.42 0.587 0.526 13.684 10.842 8.93 15.57 0.153 0.026
Panel C: FTSE 100
10 0.6 8.03 5.68 25.87 28.61 0.040 0.010 5.981 6.623 24.57 30.30 0.225 0.056
0.8 8.96 3.41 25.82 29.42 0.019 0.082 5.743 8.084 24.11 30.01 0.244 0.132
1 8.62 5.42 26.14 28.74 0.039 0.018 6.319 8.876 24.07 28.52 0.236 0.173
15 0.6 8.78 1.54 26.18 29.32 0.006 0.060 6.170 7.373 25.07 31.08 0.241 0.094
0.8 7.49 0.19 26.03 28.89 0.004 0.026 6.140 7.309 24.12 29.36 0.214 0.066
1 8.47 5.78 26.26 30.48 0.016 0.106 6.310 7.594 24.01 30.57 0.233 0.180
20 0.6 8.12 0.68 26.12 29.30 0.033 0.091 6.108 7.646 25.00 29.88 0.228 0.048
0.8 8.22 0.64 26.12 29.07 0.023 0.076 6.140 7.321 24.23 30.02 0.229 0.075
1 8.32 2.24 26.17 29.43 0.037 0.068 6.138 7.613 23.62 29.92 0.230 0.108
Panel D: UK Filter
10 0.6 14.16 2.21 18.43 23.56 1.545 0.908 11.532 5.806 22.81 29.94 0.360 0.017
0.8 12.40 7.37 18.40 23.53 1.540 1.322 11.741 8.974 22.59 30.14 0.323 0.221
1 12.91 23.42 18.47 22.11 1.506 1.353 11.692 9.453 22.38 28.14 0.333 0.560
15 0.6 14.91 5.58 18.45 23.98 1.556 0.908 11.403 4.967 23.06 28.91 0.376 0.181
0.8 14.81 7.32 18.57 23.19 1.602 1.126 12.077 6.813 22.93 30.08 0.373 0.220
1 12.13 8.57 18.59 24.84 1.560 0.947 11.759 5.099 22.40 30.45 0.317 0.245
20 0.6 15.06 8.22 18.38 24.71 1.595 1.115 11.618 6.180 22.97 29.35 0.379 0.237
0.8 14.55 6.86 18.38 24.85 1.600 0.995 11.910 5.192 22.74 30.23 0.368 0.209
1 14.03 9.44 18.48 23.93 1.611 1.037 11.846 6.240 22.36 30.48 0.357 0.265
Panel E: US Filter
10 0.6 6.16 9.29 20.51 26.77 0.303 0.650 28.721 16.322 17.51 26.39 0.187 0.129
0.8 5.70 9.06 20.64 24.56 0.246 0.018 27.642 6.268 16.95 28.30 0.177 0.127
1 6.23 1.33 20.68 24.22 0.289 0.217 29.105 7.641 17.55 29.06 0.188 0.091
15 0.6 5.12 18.48 20.57 26.87 0.252 0.104 28.952 5.516 17.48 25.97 0.165 0.317
0.8 5.47 3.41 20.63 25.42 0.200 0.165 28.577 8.188 17.38 28.33 0.172 0.008
1 5.62 8.15 20.73 24.86 0.194 0.000 28.466 6.699 17.42 27.10 0.175 0.107
20 0.6 3.91 11.69 20.58 27.18 0.289 0.154 28.874 5.360 17.46 26.41 0.141 0.178
0.8 5.27 12.30 20.65 26.32 0.206 0.287 28.549 7.590 17.31 27.99 0.168 0.193
1 5.87 6.19 20.84 26.44 0.235 0.371 28.229 11.545 17.32 29.28 0.180 0.067
An. Ret (%) An. Vol. (%) Skewness Ex. Kurtosis Correl. Info Ratio
Panel F: Indexes
SEI 3.01 48.40 0.094 2.283

Bovespa 13.21 38.04 0.026 4.875 20.09 0.185
DJIA 7.07 28.03 0.053 4.636 12.90 0.191
FTSE 100 6.01 27.42 0.009 5.374 24.34 0.182
S&P500 9.46 30.07 0.162 5.999 14.51 0.235
DJ UBS Energy-TR 18.94 36.21 0.166 1.102 43.83 0.477
Rogers Energy Commodity-TR 6.15 41.11 0.189 2.099 44.02 0.192
357
information available from the latest price data does make a differ-
ence in reducing the portfolios volatility, but the small return
improvement coupled with the rebalancing costs outweighs the
volatility benefits. Results are consistent for all cases for the risk
return trade-off l. Between monthly and quarterly rebalancing, the
differences are relatively small, but the information ratios are, in
most cases, higher for the quarterly rebalanced portfolios. Under the
buy-and-hold scenario, the best performance in terms of information
ratios is reported for the Bovespa portfolios, and under both monthly
and quarterly rebalancing this is reported for the US Filter portfolios.
In most cases, negative information ratios are reported, indicating
that these portfolios over the out-of-sample period underperform
against the benchmark, as they are associated with the lowest excess
returns.5 This observation can be explained by the fact that energy
markets, as represented by the SEI, have been resistant to the
economic recession, even although they have experienced one of the
most severe up-and-down trends in their history.
The relatively low correlations of the selected equity portfolios
with the SEI (between 9% and 31%) suggest that investors who want
to participate in the energy sector can still benefit from the addition
of the selected baskets. This observation aligns with the findings of
Buyuksahin et al (2010), that the correlation between equity and
commodity returns is not often greater than 30%. Also, correlation is
not the most appropriate performance measure, as it only measures
the degree to which the selected equity baskets and the SEI move in
tandem, and does not capture the magnitude of the returns and their
trajectories over time. Equity returns deviate from a normal distribu-
tion, displaying skewness and fat tails. The same is true for the
returns of the SEI that exhibit positive skewness and relatively high
excess kurtosis. Both futures commodity indexes have excess
kurtosis similar to the SEI, with their skewness, however, being
negative. Most equity portfolios selected by both the DE and GA
exhibit negative skewness, indicating that the equity portfolios have
more weight in the left tail of the distribution, in contrast with the
SEI, which has more weight in the right tail.
Finally, as a robustness check, a nave strategy of randomly
selected stocks has been tested, forming equally weighted portfolios
constituted of 10, 15 and 20 stocks. The stocks are selected from the
same five equity pools used by the EAs from a uniform distribution,
358
thus giving equal probability for all stocks chosen. The evidence
confirms that the strategy and methodology used in this chapter are
much more efficient and stable in achieving a good tracking perfor-
mance (low RMSEs), and good returns relative to the SEI (positive or
very small negative ERs). Under the nave strategy, there is a large
dispersion of outcomes and no consistency.6
CONCLUSIONS
In this chapter, a geometric average Spot Energy Index is constructed
and then its performance is reproduced with stock portfolios. This is
achieved by investing in small baskets of equities selected from five
stock pools: the Dow Jones, FTSE 100, Bovespa Composite and the
UK and US Filters. The investment methodology used employs two
advanced evolutionary algorithms: the GA and the DE. Both algo-
rithms are self-adaptive stochastic optimisation methods, superior to
other rival approaches when applied to the index-tracking problem.
To test the performance of the tracking baskets, three different rebal-
ancing scenarios were examined, also taking transaction costs into
consideration: buy-and-hold; monthly rebalancing; and quarterly
rebalancing. For comparison reasons, the performance of a nave
investment strategy of randomly selected stocks forming equally
weighted portfolios was also reported.
It was found that energy commodities, as proxied by the SEI, can
have equity-like returns, since they can be effectively tracked with
stock portfolios selected by the investment methodology followed
here. Overall, during the three-year period examined, which reflects
a period before, during and towards the end of the global economic
recession, an investor would realise positive returns by investing in
commodities, as the SEI returns suggest. With the methodology
employed, that performance is closely replicated and, in the case of
the energy-related stock portfolios and those selected from the
Bovespa equity pool, the benchmark index is even outperformed. In
most cases, there seem to be no major differences between the DE
and GA selected portfolios, although the GA tends to select portfo-
lios that have a lower tracking error. Both algorithms mostly do not
utilise the maximum number of stocks allowed to select, with the DE
being more stable in the number of stocks picked between the
various cases of the riskreturn trade-off; the GA tends to select port-
folios quite different in terms of their composition.
359
On average, based on the results presented here, portfolios with 15

stocks and a riskreturn trade-off value of 0.8 are the most desirable
combination providing the best results for most tracking portfolios.
Also, it was found that when rebalancing, the additional information
available from the latest price data does make a difference on
reducing the portfolios volatility; the resulting return deterioration,
however, outweighs the volatility benefits leading to smaller infor-
mation ratios. Moving from the buy-and-hold strategy to quarterly
rebalancing and then to the more frequent monthly rebalancing
strategy, returns tend to deteriorate for most selected portfolios, by
both the DE and the GA. Nonetheless, the same holds for the portfo-
lios volatilities that also tend to go down when moving from no
rebalancing to the more frequent one. Between monthly and quar-
terly rebalancing, the differences are relatively small in terms of the
portfolios return and volatility performance; however, the informa-
tion ratios are in almost all cases higher for the quarterly rebalanced
portfolios. The only exception is for the US Filter in the case of the
baskets selected by the GA. Thus, it was concluded that greater
capital efficiency can be achieved with rebalancing, preferably every
quarter, compared to the buy-and-hold strategy.
The investment approach proposed in this chapter for tracking the
performance of the energy sector with stocks selected by two innov-
ative evolutionary algorithms promotes a cost-effective
implementation and true investability. While most mutual funds
cannot invest in commodities directly, they can track the perfor-
mance of the SEI by investing in the stocks selected by the
evolutionary algorithms used here. There are many investment
houses around the globe that use evolutionary algorithms for tactical
asset management.7 The work and findings presented in this chapter
can encourage asset and fund managers to recognise the importance
of the energy sector and prompt them to set up similar funds that
will track the constructed Spot Energy Index. To that end, the
proposed methodology suggests an effective, and at the same time
least-expensive, way to operate such a fund, giving the full flexibility
of any investment style, long or short, that equities can provide.
360
APPENDIX
Differential evolution algorithm
DE is a population-based stochastic optimization algorithm that
employs mutation, recombination (crossover) and selection opera-
tors to evolve iteratively an initial set (population) of NP randomly
generated N-dimensional solutions. At each iteration (generation),
the algorithm applies the aforementioned evolutionary operators to
each one of the available solutions. In particular, let xiG denote the
solution vector i (i = 1, , NP) at a generation G, xijG be the jth element
of xiG, and x*G the best solution from generation G (specified
according to the problems objective function). Having xiG as the
starting basis, a new solution xiG+1 is constructed replacing xiG in the
next generation G + 1. The solution updating process is performed in
the following three steps:
1. A mutant solution is constructed by combining xiG with x*G

and two other randomly selected (different) solutions x and x
from the current generation: vi = xiG + F (x*G xiG) + F (x
x). The mutation constant F (0,2] controls the rate at which
the population evolves.
2. The parent solution xiG and the mutant vector vi are recom-
bined to produce a crossover solution ui, using the exponential
scheme as shown in Figure 13A.1 (for simplicity the generation
index G is not shown in the figure), where l and j* are randomly
selected from {1, 2, , N}, such that the part of ui derived from
vi is analogous to a user-defined crossover probability CR (with
higher values corresponding to a stronger impact of vi).
3. The crossover solution ui is compared against the parent vector
xi,G on the basis of the problems objective function f. If f(ui)
f(xiG), then xiG+1 is set equal to ui (ui replaces xi,G in the next
generation); otherwise, xiG+1 is set equal to xiG.
The iterative procedure terminates when a stopping criterion is met

(eg, after a predefined number of generations is explored).
Genetic algorithm
Similarly to the DE algorithm, a GA is also a population-based
stochastic optimisation process. It uses the same evolutionary opera-
tors, but implements them in a different way and does not follow the
361
Figure 13A.1 DEs exponential crossover scheme
Parent solution Parent solution

xi1, xi2, , xiN vi1, vi2, , viN
Crossover xi1 xi2 xi,l1 vil vi,l+1 vi, j*1 xij* xiN
solution
greedy approach adopted by DE. Starting with an initial (random)

population of solutions, the algorithm proceeds iteratively over a
number of generations. In the GA implemented in this chapter,
the following algorithmic steps are performed at each iteration
(generation).
1. A pair of parent solutions x and y is selected from the current

population using a tournament selection procedure. Under this
scheme, k individuals (tournament size) are randomly selected
from the population with replacement, and only the best indi-
vidual (according to the problems objective function) is
selected as a parent.
2. The parent solutions are used to perform the crossover opera-
tion with a pre-specified crossover probability (this probability
controls the frequency with which crossover is performed).
Under the arithmetic crossover scheme this operation leads to a
new pair of solutions x = rx + (1 r)y { x, y} and y = (1 r)x +
ry, where r is a random number drawn from the uniform distri-
bution in [0, 1].
3. The crossover solutions are subject to mutation. In this study
the uniform mutation strategy is employed, under which pmN
randomly selected elements of a solution vector are replaced by
random values selected uniformly from a pre-specified range.
The mutation probability pm controls the frequency of the
mutation changes.
The pair of solutions resulting from the mutation operator is placed

in the next generation of solutions, and the above three steps are
repeated until the new population is fully formulated. The proce-
dure ends as soon as a termination criterion is met (eg, the
362
Table 13A.1 Parameters of the algorithms
GA DE
Population size: Population size:

Generations: 100 Generations: 100
Crossover: arithmetic (80% probability) Mutation: rand-to-best/1 (F = 0.7)
Selection: tournament (size = 4) Crossover: exponential (CR = 0.5)
Mutation: uniform (0.5% probability)
population converges or the pre-specified number of generations is

reached).
INDUSTRY CLASSIFICATION BENCHMARK

The ICB is a company classification system developed jointly by
Dow Jones and FTSE. It is used to segregate markets into a number of
sectors within the macroeconomy. The ICB uses a system of 10
industries, partitioned into 19 super sectors, which are further
divided into 41 sectors, which then contain 114 subsectors.
The principal aim of the ICB is to categorise individual companies
into subsectors based primarily on a companys source of revenue or
where it constitutes the majority of revenue. If a company is equally
divided among several distinct subsectors, the judging panel from
both Dow Jones and FTSE makes a final decision. Firms may appeal
their classification at any time.
Table 13A.2 Industry Classification Benchmark (ICB) codes
Industry Super-sector Sector Sub-sector
0001 Oil & gas 0500 Oil & gas 0530 Oil & gas 0533 Exploration &
producers production
0537 Integrated oil & gas
0570 Oil equipment, 0573 Oil equipment &
services & distribution services
0577 Pipelines
0580 Alternative energy 0583 Renewable energy
equipment
0587 Alternative fuels
7000 Utilities 7500 Utilities 7530 Electricity 7535 Conventional electricity
7537 Alternative electricity
363
The ICB is used globally (although not universally) to divide the

market into increasingly specific categories, allowing investors to
compare industry trends between well-defined subsectors. The ICB
replaced the old classification systems used by Dow Jones and FTSE
on January 3, 2006, and it is still used by the NASDAQ, NYSE and
several other markets around the globe. All ICB sectors are repre-
sented on the New York Stock Exchange except Equity Investment
Instruments (8980) and Non-equity Investment Instruments (8990).
Table 13A.2 presents the ICB codes used for filtering all US and
UK stock markets, creating the two energy-related stock pools: the
US Filter and UK Filter, respectively.
1 Revisions can occur for a number of reasons, including additions or deletions, mergers,
splits and dividends.
2 The first listed commodity ETF was the streetTRACKS Gold Shares ETF, with its sole assets
being gold bullion and, from time to time, cash.
3 If the number of positive elements of x is smaller than K, then all positive elements of x are
used.
4 The results for both the no rebalancing and monthly rebalancing scenarios are available
upon request.
5 Note that investors who would have taken short positions on these baskets would realise
the highest excess returns.
6 The results of the naive strategy are available upon request.
7 First Quadrant, a US-based investment firm, started using EAs in 1993 to manage its invest-
ments; at the time, US$5 billion was allocated across 17 countries around the world,
claiming to have made substantial profits (Kieran, 1994).
REFERENCES
Alexander, C. and A. Dimitriu, 2002, The Cointegration Alpha: Enhanced Index

Tracking and Long-short Equity Market Neutral Strategies, ISMA Discussion Papers in
Finance, 200208, ISMA Centre, University of Reading.
Barber, B. M. and T. Odean, 2000, Trading is Hazardous to Your Wealth: The Common
Stock Investment Performance of Individual Investors, The Journal of Finance, 55(2), pp
773806.
Beasley, J. E., N. Meade and T.-J. Chang, 2003, An Evolutionary Heuristic for the Index
Tracking Problem, European Journal of Operational Research, 148(3), pp 62143.
Bodie, Z. and V. Rosansky, 1980, Risk and Return in Commodity Futures, Financial
Analyst Journal, 36(3), pp 2739.
Buyuksahin, B., M. S. Haigh and M. Robe, 2010, Commodities and Equities: A Market
of One?, Journal of Alternative Investments, 12(3), pp 7695.
364
Canakgoz, N. A. and J. E. Beasley, 2008, Mixed-integer Programming Approaches for

Index Tracking and Enhanced Indexation, European Journal of Operational Research, 196, pp
38499.
Chang, T.-J., S.-C. Yang and K.-J. Chang, 2009, Portfolio Optimization Problems in
Different Risk Measures Using Genetic Algorithm, Expert Systems with Applications, 36(7),
pp 10,52937.
Chen, C. and R. H. Kwon, 2012, Robust Portfolio Selection for Index Tracking,
Computers & Operations Research, 39, pp 82937.
Commodity Futures Trading Commission, 2008, Staff Report on Commodity Swap

Dealers and Index Traders with Commission Recommendations, September, pp 170.
Dunis, C. L. and R. Ho, 2005, Cointegration Portfolios of European Equities for Index
Tracking and Market Neutral Strategies, Journal of Asset Management, 6(1), pp 3352.
Erb, C. B. and C. R. Harvey, 2006, The Strategic and Tactical Value of Commodity
Futures, Financial Analyst Journal, 62(2), pp 6997.
Fama, E. F. and K. R. French, 1987, Commodity Futures Prices: Some Evidence on

Forecast Power, Premiums and the Theory of Storage, Journal of Business, 60(1), pp 5573.
Feoktistov, V. and S. Janaqi, 2004, Generalization of the Strategies in Differential

Evolution, presented at the 18th International Parallel and Distributed Processing
Symposium, Santa Fe, New Mexico, US.
Frino, A. and D. R. Gallagher, 2001, Tracking S&P 500 Index Funds, The Journal of
Portfolio Management, 28(1) pp 4455.
Gaivoronski, A. A., S. Krylov and N. Van der Wijst, 2004, Optimal Portfolio Selection
and Dynamic Benchmark Tracking, European Journal of Operational Research, 163(1), pp
11531.
Goldberg, D. E., 1989, Genetic Algorithms in Search, Optimization, and Machine Learning
(Reading, MA: Addison-Wesley).
Gordon, R., 2006, Commodities in an Asset Allocation Context, Journal of Taxation of

Investments, 23(2), pp 18189.
Gorton, G. and G. Rouwenhorst, 2006, Facts and Fantasies about Commodity Futures,
Financial Analyst Journal, 62(2), pp 4768.
Krink, T., S. Mittnik and S. Paterlini, 2009, Differential Evolution and Combinatorial
Search for Constrained Index-tracking, Annals of Operations Research, 172, pp 15376.
Krink, T. and S. Paterlini, 2009, Multiobjective Optimization Using Differential

Evolution for Real-world Portfolio Optimization, Computational Management Science, 8, pp
15779.
Konno, H. and T. Hatagi, 2005, Index-plus-alpha Tracking Under Concave Transaction

Cost, Journal of Industrial and Management Optimisation, 1(1), pp 8798.
Larsen-Jr, G. A. and B. G. Resnick, 1998, Empirical Insights on Indexing, The Journal of

Portfolio Management, 25(1), pp 5160.
Li, Q., L. Sun and L. Bao, 2011, Enhanced Index Tracking Based on Multi-objective
Immune Algorithm, Expert Systems with Applications, 38, pp 6,10106.
Malkiel, B., 1995, Returns from Investing in Equity Mutual Funds, 1971 to 1991, The
Journal of Finance, 50(2), pp 54972.
365
Maringer, D., 2008, Constrained Index Tracking Under Loss Aversion Using Differential
Evolution, in A. Brabazon and M. ONeil (Eds), Natural Computing in Computational
Finance, Studies in Computational Intelligence (Berlin: Springer): pp 724.
Maringer, D. and O. Oyewumi, 2007, Index Tracking Constrained Portfolios, Intelligent

Systems in Accounting, Finance and Management, 15, pp 5771.
Markowitz, H., 1952, Portfolio Selection, The Journal of Finance, 7(1), pp 7791.
Nathan, R., 2004, A Review of Commodity Indexes, Journal of Indexes, June/July, pp

3035.
Oh, K. J., T. Y. Kim and S. Min, 2005, Using Genetic Algorithm to Support Portfolio
Optimization for Index Fund Management, Expert Systems with Applications, 28, pp
37179.
Plaxco, L. M. and R. D. Arnott, 2002, Rebalancing a Global Policy Benchmark, Journal of

Portfolio Management, 28(2), pp 922.
Price, K. V., R. M. Storn and J. A. Lampinen, 2005, Differential Evolution: A Practical

Approach to Global Optimization (Heidelberg: Springer).
Rohweder, H. C., 1998, Implementing Stock Selection Ideas: Does Tracking Error
Optimization Do Any Good?, The Journal of Portfolio Management, 24(3), pp 4959.
Ruiz-Torrubiano, R and A. Surez, 2009, A Hybrid Optimization Approach to Index

Tracking, Annals of Operations Research, 166, pp 5771.
Sarker, R. and H. Abbass, 2004, Differential Evolution for Solving Multi-objective

Optimization Problems, Asia-Pacific Journal of Operations Research, 21(2), pp 22540.
Sarker, R., K. Liang and C. Newton, 2002, A New Multi-objective Evolutionary

Algorithm, European Journal of Operational Research, 140(1), pp 1223.
Scozzari, A., F. Tardella, S., Paterlini and T. Krink, 2012, Exact and Heuristic
Approaches for the Index Tracking Problem with UCITS Constraints, Center for
Economic Research (RECent) 081, University of Modena and Reggio E., Dept. of
Economics, Italy.
Sharpe, W., 1991, The Arithmetic of Active Management, The Financial Analyst Journal,
47(1), pp 79.
Soleimani, H., H. R. Golmakani and M. H. Salimi, 2009, Markowitz-based Portfolio

Selection with Minimum Transaction Lots, Cardinality Constraints and Regarding Sector
Capitalization Using Genetic Algorithm, Expert Systems with Applications, 36(3), pp
5,05863.
Sorenson, E. H., K. L. Miller and V. Samak, 1998, Allocating Between Active and Passive
Management, Financial Analyst Journal, 54(5), pp 1831.
Stopford, M., 2009, Maritime Economics (3e) (London: Routledge).
Storn, R. and K. Price, 1995, Differential Evolution: A Simple and Efficient Adaptive
Scheme for Global Optimization Over Continuous Spaces, Technical Report TR-95012,
International Computer Science Institute, Berkeley.
Storn, R. and K. Price, 1997, Differential Evolution, A Simple and Efficient Heuristic
Strategy for Global Optimization Over Continuous Spaces, Journal of Global Optimization,
11(4), pp 34159.
366
Stoyan, S. J. and R. H. Kwon, 2010, A Two-stage Stochastic Mixed-integer Programming

Approach to the Index Tracking Problem, Optimization and Engineering, 11(2), pp 24775.
Sullivan, R., A. Timmermann and H. White, 1999, Data-snooping, Technical Trading

Rule Performance, and the Bootstrap, The Journal of Finance, 54(5), pp 1,64791.
Tang, K. and W. Xiong, 2009, Index and the Financialization of Commodities, working
paper, September.
Wang, M. Y., 1999, Multiple-benchmark and Multiple-portfolio Optimization, Financial

Analyst Journal, 55(1), pp 6372.
White, H., 2000, A Reality Check for Data Snooping, Econometrica, 68(5), pp 1,097126.
Woodside-Oriakhi, W., C. Lucas and J. E. Beasley, 2011, Heuristic Algorithms for the
Cardinality Constrained Efficient Frontier, European Journal of Operational Research, 213(3),
pp 53850.
Zhang, J. and D. Maringer, 2010, Index Mutual Fund Replication, in A. Brabazon, M.

ONeil and D. Maringer (Eds), Natural Computing in Computational Finance, Studies in
Computational Intelligence (Berlin: Springer): pp 10930.
367
Part III
Market Developments and

Risk Management
14
Enterprise Risk Management for
Energy and Commodity Physical and
Financial Portfolios
Carlos Blanco
NQuantX LLC and MTG Capital Management
This chapter will present an enterprise risk management (ERM)

framework for energy and commodity physical and financial portfo-
lios based on the three usual building blocks of policies and
governance, methodologies and metrics, and infrastructure.1 The
framework can be used to structure as well as conduct due diligence
on the soundness of the risk-management process for all material
risks such as market, credit, operational and liquidity risk as well
as their interactions.
This chapter is divided according to these blocks, with the first
section examining policies and governance, and the need to integrate
risk management in the governance structure of the firm. We then
discuss methodologies and metrics, particularly valuation, risk and
performance metrics for physical and derivatives portfolios, before
moving on to infrastructure, and delving into people, data, opera-
tions and systems.
POLICIES AND GOVERNANCE

A risk governance framework integrates risk management into the
governance structure of the firm to ensure that risk groups have the
independence, stature and adequate resources to fulfill their respon-
sibilities within the overall business strategy of the firm.
Over the years, many risk management groups that were believed
371
to follow best practices have been repeatedly unable or unwilling

to prevent their institutions from engaging in excessive risk taking,
which eventually resulted in heavy losses that bankrupted those
firms. Some of the key reasons for this are the assymetric compensa-
tion structures at trading desks that encourage excessive short-term
risk taking and the lack of stature of the risk groups compared to the
revenue generating units, as well as the willingness of management
teams and boards to turn a blind eye when profits are rolling in.
Board and senior management teams have the responsibility to
manage the main risks of the firm. However, few board members
have a strong background in financial risk management and few risk
managers have the breadth of skills and experience required to
interact directly with board members and understand or influence
the firms strategy and the process that sets the risk appetite and
associated boundaries.
The large exposures accumulated in the real estate and credit
markets at large financial institutions such as Bear Stearns, Lehman
Brothers, Merrill Lynch and AIG before the global financial crisis of
200708 caught many boards, senior management teams and risk
groups by surprise. The governance structure in those firms ulti-
mately failed to provide the oversight and early warning signals that
would have prevented firms from taking on too much risk in certain
areas.
In other cases, even although risk managers informed senior
management about the magnitude of the risks taken and the poten-
tial catastrophic consequences for their firms, senior-level executives
decided to sugarcoat it for their boards, or omitted critical details.
Lehman Brothers, AIG, Bear Stearns and BP are painful examples of
risk groups lack of independence and inability to communicate the
firms risk all the way to board level. Another example is BPs lack of
preparedness and its incompetent response to the oil drilling plat-
form explosion and subsequent oil spill in the Gulf of Mexico in 2008,
which has become a case study on crisis mismanagement.
Boards can also ensure that the risk management group roles and
responsibilities are aligned with value creation (or at least the preven-
tion of value destruction). Unless the risk management efforts are
structured in the right context, risk management activities will take a
secondary role, and may ultimately end up destroying value and
negatively interfering with the business strategy of the firm.
372
ENTERPRISE RISK MANAGEMENT FOR ENERGY AND COMMODITY PORTFOLIOS
An example of best practices in the integration of risk manage-

ment within the overall governance structure is the risk management
policy of BHP Billiton PLC, one of the largest diversified resources
companies in the world (see Panel 14.1).
PANEL 14.1: BHP BILLITON RISK MANAGEMENT POLICY2

BHP Billitons risk management policy (see below) defines the groups
approach to risk management, linkage to the corporate objective and inte-
gration into its business processes.
Risk is inherent in our business. The identification and management

of risk is central to delivering on the corporate objective.
Risk will manifest itself in many forms and has the potential to impact
the health and safety, environment, community, reputation, regula-
tory, operational, market and financial performance of the group and,
thereby, the achievement of the corporate objective.
By understanding and managing risk we provide greater certainty and
confidence for our shareholders, employees, customers and
suppliers, and for the communities in which we operate.
Successful risk management can be a source of competitive
advantage.
Risks faced by the Group shall be managed on an enterprise-wide
basis. The natural diversification in the Groups portfolio of
commodities, geographies, currencies, assets and liabilities is a key
element in our risk management approach.
We will use our risk management capabilities to maximise the value
from our assets, projects and other business opportunities and to
assist us in encouraging enterprise and innovation.
Risk management will be embedded into our critical business activi-
ties, functions and processes. Risk understanding and our tolerance
for risk will be key considerations in our decision-making.
Risk issues will be identified, analysed and ranked in a consistent
manner. Common systems and methodologies will be used.
Risk controls will be designed and implemented to reasonably assure
the achievement of our corporate objective.
The effectiveness of these controls will be systematically reviewed
and, where necessary, improved.
Risk management performance will be monitored, reviewed and
reported. Oversight of the effectiveness of our risk management
processes will provide assurance to executive management, the
board and shareholders.
The effective management of risk is vital to the continued growth and
success of our Group.
Source: BHP Billiton
373
Any investment and trading operation should clearly articulate

and communicate the core investment strategies and the firms risk
tolerance. Written policies can establish the link between its business
strategy and its tolerance for risk.
A final component of the policy dimension is the existence of clear
lines of authority and an appropriate level of risk disclosure, which
makes the risk transparent to internal and external stakeholders. The
degree of authority and independence of the risk group is a function
of the relative stature of the group within the firm. For example, a
risk group that reports directly to the heads of the business unit in
charge of revenue generation is unlikely to have the independence
and authority to prevent excessive risk taking.
It is important not to confuse detailed investment strategy and
position-level disclosures (often considered highly proprietary by
portfolio managers) with risk disclosures designed to provide assur-
ances that the risk levels are within the parameters expected by
investors. For example, an investment manager that provides value-
at-risk (VaR) disclosures to investors is not giving away proprietary
information that could be used by counterparties against the firms
portfolio.
VALUATION AND RISK METHODOLOGIES AND METRICS

The second building block of the risk process consists of the method-
ologies and metrics used to measure and manage risk, as well as their
integration in risk-adjusted performance measures.
Given the continued high volatility and extreme moves in the
energy and financial markets, one of the main contributions of the
risk groups is the calculation of key risk metrics that can assist
decision-makers with the process of identifying, measuring and
managing the firms material risks.
Since the first generation financial risk models were published in
the mid-1990s, there have been significant developments in the
modelling of market, credit, liquidity and operational risk of energy
and commodity portfolios. However, the excitement and quick
progress of the early years has now gone, and change tends to be
slower and incremental and driven by regulatory and external pres-
sures.
374
At-risk metrics: Cashflow at risk, VaR and earnings at risk

Below are listed the most commonly used market risk metrics for
energy and commodity portfolios.
VaR is a measure of the potential variability in the mark-to-

nce level and time
rm risk metric for
ntial variability of
Table 14.1 Differences between VaR, CFaR and EaR
VaR Collateral at risk EaR

(CaR) and CFaR
Market scenarios YES YES YES

Mark to market/mark to model YES YES YES
Multiple time steps (periods) NO YES YES
Portfolio ageing and walk-forward analysis NO YES YES
Netting agreements NO YES YES
Collateral and margin clauses NO YES YES
Portfolio trading/hedging strategies NO YES YES
Counterparty default NO YES YES
Hedge effectiveness rules NO NO YES
Rating downgrade NO YES YES
Operational risks NO YES YES
Dynamic hedging strategy NO YES YES
Volumetric risks NO YES YES
Source: NQuantX
375
calculations involve modelling costs and revenues related to the

operation of physical assets (for example, generation, storage), spot
purchases and sales in the spot market, as well as profit and losses
from the hedging and trading portfolio.
The process for calculating CFaR and EaR requires careful
analysis and understanding of each portfolios material risks. For
example, volume-related variability embedded in many physical
and derivative contracts and operations-related constraints such as
ramp-up and ramp-down rates and plant outages need to be
explicitly accounted for to obtain a realistic value and risk estimates.
The same is true for critical operating constraints from assets and the
material clauses in contracts, as well as the limitations of physical
and financial arbitrage strategies such as market liquidity and avail-
able hedging instruments. Figure 14.1 shows some of the
components required to compute CFaR and EaR.
Failure to incorporate material risks when evaluating a hedging or
trading strategy such as CFaR and collateral implications may have
unintended consequences and expose the firm to unwanted risks.
Figure 14.1 Cashflow-at-risk models require the integration of multiple source of

risk and portfolio components
MtM changes and

volumetric variability
(market risk)
Collateral changes CFaR

(MtM-based, AR/AP, and
premiums, downgrades) EaR
Counterparty losses
(MtM-based, netting guarantees,
AR/AP, collateral)
Source: NQuantX LLC
376
When energy and commodity prices collapsed in the summer of

2008, many firms experienced a large volume of collateral calls that
forced them to go to the capital markets to raise extra capital at a time
when banks were not extending credit to their counterparties. Firms
without contingency plans in place that had funding problems
ended up having to pay exorbitant borrowing costs or were just
unable to continue funding their hedges. For example, airlines that
had hedging programmes in place at the time accumulated large
mark-to-market losses in their hedge portfolios when crude oil and
jet fuel prices fell over 50% in just a few months in 2008. In addition,
the global financial crisis caused a sharp drop in business travel
worldwide that impacted their operating revenues. Rating agencies
downgraded many airlines due to the worsening liquidity picture
caused by the combination of lower forecasted revenues and the
large cash outflows due to collateral calls from their existing hedges.
As a result, hedging costs increased considerably at a time when
financial institutions were attempting to reduce their credit risk
exposures. The response by many airlines was to discontinue or
reduce the size of their hedging programmes.
Forward-looking key risk indicators such as CaR can measure the
maximum collateral outflows for a given confidence level, taking
into account initial and variation margin requirements for over-the-
counter (OTC)-cleared and exchange-traded contracts, as well as the
material margin clauses such as the credit support terms for OTC
transactions (eg, thresholds, independent amounts, downgrade trig-
gers, eligible collateral).
Figure 14.2 shows the potential future collateral outflows for a
derivatives portfolio, as well as the potential counterparty future
exposures as a function of simulated market environments. Both
metrics in Figure 14.2 are calculated for a 95% confidence level and
provide an indication of the potential magnitude of unsecured credit
exposures and potential margin payments.
Stress tests
In order to manage extreme event risk, the risk process should also
include stress tests that question the model assumptions and also the
market consensus view at any given moment. Stress tests are
particularly relevant in markets that experience large sudden fluctu-
ations as well as regime changes, and the results from the analysis
377
Figure 14.2 Potential future exposure and potential collateral requirement report
US$30,000,000
US$20,000,000
Potential future exposure at the

95% confidence level
US$10,000,000
US$0
US$10,000,000 Walk forward collateral

requirements
US$20,000,000 Potential collateral outflows at the

95% confidence level
US$30,000,000
4/1/2010 5/1/2010 6/1/2010 7/1/2010 8/1/2010 9/1/2010 10/1/2010 11/1/2010 12/1/2010 1/1/2011 2/1/2011 3/1/2011 4/1/2011
Collateral walk forward 95% Collateral at risk profile 95% Potential future exposure
Source: NQuantX LLC
378
can provide key insights to portfolio managers to develop contin-

gency plans.
Stress test committees that have representatives from the main
groups in the firm such as fund managers, risk managers and
analysts can proactively identify scenarios that would prove useful
preludes to market crisis and feed that information into strategic
planning, capital allocation, hedging and other major decisions.
If stress test results (as illustrated in Figure 14.3) indicate that the
hedge funds losses are beyond their tolerance level or the available
capital, then immediate instructions could be sent to the fund
managers to reduce the exposure to such event or increase its capital.
Another area where stress tests are critical is liquidity risk
management and capital adequacy. Liquidity risk management is
often mistaken for crisis management, as lack of planning often
forces companies to address liquidity risk management issues only
Figure 14.3 Stress-test results for price and volatility changes broken down by
desk
US$100,000
US$50,000
US$0
P&L (thousands)
-US$50,000
-US$100,000
-US$150,000
-US$200,000
-US$250,000 Power
-US$300,000 Crude oil and products
Agricultural
0%
%
5%
0%
5
-5
0%
-2
0%
+2
e
+5
e
ic
-2
ic
+2
e
Pr
e
ic
Pr
y
ic
t
Pr
ity
i
Pr
til
til
la
la
Vo
Vo
Agricultural Metals Crude oil and products

Gas Power Portfolio
Source: NQuantX LLC
379
when crises occur. In order to estimate liquidity risk, companies

need a framework that explicitly addresses the potential demand
and supply of cash. This framework should address medium- and
long-term horizons as well as the cost of liquidating positions in a
stressed market environment.
Risk managers can use scenario analysis and reverse stress tests to
identify scenarios that could result in liquidity shortfalls, particularly
under stress market conditions. Funding liquidity risk measurement
requires metrics such as CFaR, CaR and margin-at-risk, which iden-
tify potential margin and collateral calls in stress situations.
The most sophisticated risk and stress test models incorporate
more realistic assumptions about market dynamics, particularly in
properly modelling the tails of the distribution, including a dynamic
approach to correlation to allow firms to incorporate credit and
liquidity considerations, and finally a method to anticipate the risk
takers response to various market events. A set of principles to
measure and manage tail risk is presented in Panel 14.2.
Backtesting
Backtesting a risk model consists of evaluating whether the risk
model forecasts are adequately capturing the magnitude and
frequency of profit and losses (P&Ls). There is a wide range of quan-
titative and qualitative backtests,3 but most individual tests have low
statistical significance. As a result, the most common way to perform
backtesting is by analysing a chart with P&L series and VaR fore-
casts. The backtest procedure consists of comparing the daily VaR
with the subsequent P&L for T+1 as the VaR forecast attempts to
determine the magnitude of future P&L.
The most common VaR backtests analyse whether the number
and magnitude of exceptions is within the VaR model predictions.
Loss exceptions are those losses greater than the prior day VaR,
while gain exceptions are gain larger than the prior day VaR on the
positive tail of the distribution (VaR+). Many firms exclusively focus
on loss exceptions and ignore gain exceptions, but that may lead to
situations where large gains could go unexplained for long periods
of time and eventually turn into large unexpected losses.
An additional test consists of checking whether those exceptions
are autocorrelated, which would result in many exceptions taking
place during short periods of time that potentially could result in
380
PANEL 14.2: TAIL RISK MANAGEMENT PRINCIPLES

Risk management is ultimately the art of managing risk based on the pres-
ence of imperfect (and constantly evolving) information. Measuring and
managing risks in the tail of the distribution is both an art and a science. A
few basic risk management principles for extreme events will now be
examined.
Include all plausible scenarios of material risk factors in the analysis

Stress scenarios should integrate all material risk factors, such as market
risks (eg, price, basis, volatility), counterparty risks, and also relevant
funding and market liquidity risks in a coherent fashion. Ignoring key risks
may result in risk information offering a simplistic and inaccurate view of
tail risk that can give a false sense of security.
Choose appropriate tail risk metrics and modelling horizons

The most common risk metrics used, such as VaR and standard deviation,
fail to capture the dynamics of the tail of the distribution of potential
outcomes. Fortunately, there are other metrics, such as stress test results,
expected tail loss (ETL) and expected shortfall (ES), spectral risk measures
and probable maximum loss (PML), that can complement a risk limit
structure (see Chapter 2 of Dowd, 2005).
Measurement is just the starting point

Adequate preparation for any future crises requires forward-looking and
creative thinking, as well as carefully designed contingency plans.
Designing and conducting realistic stress tests that provide insights into
the likely portfolio gains and losses under particular extreme events is
necessary but not sufficient. The development of contingency plans to
respond to those hypothetical extreme events have lagged behind as most
firms have been shown to be ill-prepared to respond to crises.
Expect the unexpected: Account for model risk

While crises in financial, credit and energy markets often share some simi-
larities with prior ones, each new crisis has differentiating elements that
tend to catch most players by surprise. The lesson is that any contingency
plan against extreme events should leave a significant buffer to account for
variations from expected extreme scenarios.
cumulative losses or gains considerably higher than the risk model

forecasts. As a general rule, any exceptions should be investigated by
the market risk management group, and the reasons for the excep-
tion should be recorded to identify regular culprits and corrective
action taken if necessary. If the backtests show that the risk models
381
ultimately failed to capture actual P&L variability, they should be

replaced as the primary market risk control tool.
Valuation models for physical assets, contracts and financial

derivatives
Energy and commodity markets are dominated by asset-based
traders that optimise their strategies around physical assets such as
storage facilities, pipelines and power plants. The traders and opera-
tors of those assets attempt to maximise risk-adjusted profits based
on observable market spreads and the specific asset operating
constraints. The valuation, risk metrics and hedging ratios calculated
from a static model are not just likely to be inaccurate, but could lead
to suboptimal decisions that would impact the profitability.
In order to capture the multiple risk dimensions involved in
hedging and trading, a dynamic simulation framework with three
critical components is needed: the ability to handle multiple risk
factors (eg, price risk, credit events, operational issues), multiple
instruments (eg, physical contracts, derivatives) and the ability to
capture events taking place at multiple steps in time.
Dynamic risk simulation involves modelling the variability of one
or more metrics (eg, cash flow, earnings, mark-to-market, liquidity)
based on a realistic evolution of a set of key state variables, as well as
the firms response to those changes (eg, operating, hedging and
trading strategies). The analysis consists on leaping forward in
time by simulating risk variables at various point in time in the
future and evaluating a series of value and risk metrics (eg, costs,
revenues, profits, VaR) under each of those scenarios.
An added benefit of using dynamic simulation-based risk tools is
the potential for risk management to play a larger role in strategic
business decisions at various levels of the firm. For example, simula-
tion analysis can assist trading and operating groups in developing
asset optimisation and hedging strategies based on the evaluation of
riskreturn trade-offs. It can also help finance groups creating
forward-looking earnings projections by ensuring they are consis-
tent with the risk appetite of the firm. Another critical area is the
evaluation of important investment and divestment decisions from a
marginal and stand-alone risk point of view.
Advances in financial engineering and computational finance
such as least squares Monte Carlo and dynamic programming have
382
allowed for the widespread use of dynamic risk simulation solutions

in energy firms.
Risk-adjusted performance measurement

Identifying, measuring, managing and pricing risk involves
designing appropriate policies and systems so different business
units and risk-taking activities can be evaluated with a risk-adjusted
return in mind. The practical requirements to implement such
changes involve modifying performance measurement mechanisms
to integrate the risk and return numbers in the bonus allocation
process, as well as changing the risk systems, to allow for risk
numbers calculations at the risk-taking unit level.
Risk-adjusted return on capital (RAROC) measures, which are
widely used in the financial services industry, provide a common
measurement unit for risk-adjusted returns on allocated (ex ante) and
utilised (ex post) risk capital.
A RAROC system can assist managers to determine the most effi-
cient generators of revenue on a risk-adjusted basis, as well as set the
threshold returns given the risk assumed to generate them. Let us
assume that we are the trading manager of a commodity trading
desk and have two traders. Both of them made a profit of US$10
million, but on average one of the traders used 80% less risk capital
than the other. If the trading manager only considers the size of the
gains, both should get a similar bonus. However, from a risk-
adjusted perspective, the trader that took less risk should receive a
higher bonus.
An investment evaluation process based on economic capital
considerations, where decisions are based on a risk-adjusted return
basis, encourages corporate managers to become risk managers
because they must take risk into consideration when allocating
resources internally and making investment and divestment deci-
sions. Determining the economic capital allocated to each activity or
business unit provides senior management with a mechanism to link
risk and return, and therefore provide a riskreward signal that can
be used at different levels of the firm.
INFRASTRUCTURE
The third building block is the risk infrastructure. The infrastructure
subcomponents are: people, systems, data and operations.
383
Risk management functions have too often been built around

quantitative risk managers who lacked market experience and requi-
site managerial skills. A chief risk officer (CRO) with the right skills,
experience, independence and courage to perform the job is a critical
component in the risk management organisation. The CROs effec-
tiveness is a direct function of the power and independence granted
that position, the quality of the people in the risk organisation, the
associated culture, and incentives and experience. If we were to boil
down empowerment to one question, then we might ask an organi-
sation if its CRO is empowered to be proactive, as opposed to
remaining reactive.
The budget of the overall risk management function determines
the scope and depth of activities that can be performed. The educa-
tion and experience of the risk management personnel is a direct
reflection of an organisations ability to hire and retain first-class risk
managers. The stature of risk managers is also important; if risk
management personnel do not have the stature to be able to stand up
for their beliefs under pressure, it is a recipe for eventual failure.
Risk groups must balance the day-to-day tactical issues such as
risk measurement, reporting and limit checking with the more
strategic aspects of evaluating business decisions that have a mate-
rial impact on the firms risk profile and whose effect may only be felt
further into the future.
Risk management systems and data

Financial risk management and technology advances have made
possible the integration of data from multiple sources in order to
provide the firmwide perspective required to forecast risk scenarios
involving multiple risk dimensions.
However, many energy trading and risk systems have failed to
keep up to date with the advances in risk analytics. Those systems
excel at performing tasks such as scheduling, nomination or
accounting of physical and financial trades, but offer limited risk
functionality and lag behind in their ability to perform sophisticated
risk analysis.
Many of the pioneering risk software firms that greatly
contributed to important methodological advances in energy risk
modelling are now part of larger software and consulting firms.
Experience has painfully shown that their ability to innovate beyond
384
PANEL 14.3: RISK MANAGEMENT LESSONS FROM OSPRAIE

CAPITAL MANAGEMENT4
Ospraie Management LLC, one of the largest commodity trading funds in
the world, was forced to liquidate its largest fund after losing 38.6% in the
first three quarters of 2008.
The investment firm was run by Dwight Anderson, an experienced
commodities trader with an excellent trade record up to that point. In a
letter to investors after closing the fund, Anderson wrote that, I am
extremely disappointed with this result and the funds sudden reversal in
performance. After nine years of striving to be a good steward of your
capital, I am very sorry for this outcome.
Just a few months before closing the fund because of the large losses,
Anderson told Bloomberg We do everything that we can to manage the
risk, and I think were better at it today than we were a year ago.
The sudden reversal in the funds performance caught most investors by
surprise. For example, in 2007 the head of Credit Suisses New York-based
alternative investments group, which managed US$134 billion in private
equity and hedge fund assets, said that Andersons the best-in-class
player in dealing in the world of basic industries and commodities.
pure technology and computational solutions has come to a near

halt. For example, most valuation and risk models are still based on
static portfolios over short time periods that ignore material risks
such as volumetric and operational risks. In addition, most models
used by risk practitioners still assume that market changes are
lognormal, which fail to account for key characteristics of energy and
commodity prices such as mean reversion and jumps.5
One of the major gaps that exists in the risk management process
at many firms is the lack of effective communication between risk
groups and the senior management team. A poorly informed
management exposes the organisation to risk blind spots.
Some risk managers have taken a proactive role, and regularly
identify those key risk information gaps (see Table 14.3) and continu-
ally develop and improve the tools to breach them.
SUMMARY AND CONCLUSIONS

Energy and commodity markets are some of the most volatile
markets in the world, and firms operating in those markets should
approach risk management with caution.
Firms that implement a rigorous enterprise risk management
385
Table 14.3 Main information gaps from a risk management perspective
Organisational Risk management information gaps Reports and metrics

level
Board of Greater transparency on material risks and Risk and hedging strategy
directors better understanding of high-level risk dashboards.
return trade-offs of alternative hedging
alternatives.
Senior Knowledge of firmwide exposures and Firmwide exposure and
management interactions. Impact of hedging on at-risk reports.
shareholder value maximisation. Hedge recommendations
Evaluation of riskreturn trade-offs and Stress-test reports.
optimisation based on risk tolerance and
multiple constraints.
CFO/Treasury Anticipate potential cashflow shortfalls and CFaR; CaR; hedge

develop contingency plans. effectiveness; EaR.
Evaluation of pre- and post-hedge
effectiveness. Negotiation of key price and
collateral clauses in long-term contracts.
Risk-adjusted pricing for large transactions.
Procurement/ Assistance with (re)negotiation of critical Cost-at-risk; operating

logistics groups contract price and volume-related clauses. cashflow reports.
Increased focus on operational efficiency
around physical procurement contracts.
Benchmarks to determine group
performance.
Market risk In-depth understanding of multiple risk Dynamic risk simulation of
managers dimensions before and after hedging at the material risks. Valuation
portfolio level (market, collateral, liquidity, and risk adjustments.
cashflow).
Credit risk Dynamic counterparty risk assessments. Potential future exposure
managers Impact of netting and collateral clauses in and credit risk reports. Risk
OTC master agreements (netting and charges and risk-adjusted
collateral). pricing.
Integration of accounts payable and
receivables, as well as potential collateral
needs in the cashflow management
programme.
Source: NQuantX LLC
process can develop a competitive advantage that will allow them to

weather the storm of adverse market conditions. Being aware of best
practices and striving to implement them are therefore key not just to
success, but to having good prospects for longer-term survival.
A risk management process is as effective as its weakest link, and
386
therefore it is important to ensure that all the elements of the process

are robust and integrated. The framework based on policies and
governance, methodologies and infrastructure introduced in this
chapter can assist energy and commodity market participants design
a robust and comprehensive risk management process.
1 For more information, see Blanco and Mark (2004).

2 BHP Billiton Risk Management Policy (see http://www.bhpbilliton.com/home/aboutus/
ourcompany/Documents/Risk%20Management%20Policy.pdf).
3 There is an excellent overview of backtests in Dowd (2005).
4 Burton, K., S. Kishan and C. Harper, 2008, Ospraie to Close Flagship Hedge Fund After
38% Loss, Bloomberg, September 3.
5 For readers interested in a more detailed discussion of energy and commodity spot and
forward price models, see Blanco and Pierce (2012).
REFERENCES
Aragons, J. R., C. Blanco, K. Dowd and R. Mark, 2006, Market Risk Measurement and
Management for Energy Firms, in P. C. Fusaro (Ed), Professional Risk Managers Guide to
Energy and Environmental Markets (Wilmington, DE: PRMIA Publications): pp. 6982.
Blanco, C. and M. Pierce, 2012, Spot Price Process for Energy Risk Management, Energy
Risk, March.
Blanco, C. and M. Pierce, 2012, Multi-factor Forward Curve Models for Energy Risk
Management, Energy Risk, April.
Blanco, C. and R. Mark, 2004, ERM for Energy Trading Firms: ERM Starts with Risk
Literacy, Commodities Now, September, pp 7882.
Blanco, C., 2010, Collateral, Cash Flow and Earnings at Risk, WorldPower, Isherwood
Publications.
Blanco, C. and M. Pierce, 2010, Integrated Risk Modeling for Trading and Hedging
Decisions, WorldPower, Isherwood Publications.
Dowd, K., 2005, Measuring Market Risk (2e) (Hoboken, NJ: Wiley).
387
15
Credit Valuation Adjustment (CVA) for
Energy and Commodity Derivatives
Carlos Blanco and Michael Pierce
NQuantX LLC and MTG Capital Management; NQuantX LLC
Traditionally, counterparty and liquidity risks have been largely

ignored in the valuation and risk measurement of energy and
commodity portfolios. The main reasons for this were the lack of
commonly accepted methodologies to measure and price those risks,
as well as the general perception that they were relatively immate-
rial. However, large credit losses and funding liquidity problems,
significant advances in credit risk measurement technology and
changes in accounting standards and regulations such as the Dodd
Frank Act have led to an increased focus on improving counterparty
and liquidity risk management practices.
At the centre of the credit revolution is the concept of credit valua-
tion adjustment (CVA), which likely to play as great a role for credit
risk management as value-at-risk (VaR) did for the practice of
market risk management. In this chapter, we will examine the
concept of CVA and show how to calculate CVA at the trade and
portfolio levels. We will also discuss the allocation of portfolio CVA
and active credit risk management with CVA desks, as well as how
to set up a system of credit risk charges.
CVA IN A NUTSHELL
In simple terms, CVA is the price of credit risk for a deal or portfolio
with a given counterparty. When two entities enter into a derivative
transaction, they also exchange an implicit option to default.
389
From a valuation perspective, the fair value measurements of

derivative contracts should include a risk adjustment reflecting the
amount market participants would demand because of the credit
risk in the future cashflows that will be exchanged through the life of
the contract. The fair value of the embedded default option that
could result in a credit loss is the CVA.
Calculating CVA requires estimating the difference between the
mark-to-market discounting expected cashflows using the risk-free
rate and the credit-adjusted mark-to-market, which consists of incor-
porating the credit risk of the transaction into the mark-to-market
calculations.
CVA = Mark-to-market (risk-free) Mark-to-market (credit-adjusted)
CVA and debt valuation adjustment

CVA can be unilateral or bilateral, depending on whether the adjust-
ments are based on one or two of the parties in the deal. In unilateral
adjustments, the entity performing the CVA calculations only
discounts the derivatives assets (eg, in the money transactions) using
the credit-adjusted curves of the counterparties, while no adjust-
ments are made for liabilities (out-of-the-money transactions). In
addition, the entity performing the calculations could also perform a
similar adjustment for its liabilities, but using its own credit risk-
adjusted curves. These adjustments are also known as the debt
valuation adjustment (DVA).
In bilateral adjustments, the entity performing the calculations
takes into account the effect of the counterpartys credit risk in deter-
mining the prices they would receive to transfer an asset, as well as
the effect of the entitys credit risk in determining the prices they
would pay to settle that liability. The main differences between bilat-
eral and unilateral CVA calculations is that the former integrates the
credit risk from the two parties in the transaction, and therefore is
calculated as the net difference between the unilateral CVA and the
DVA.
Another risk adjustment is the funding valuation adjustment
(FVA), which incorporates funding costs for the position such as
initial and variation margin, and collateral.
390
CREDIT VALUATION ADJUSTMENT (CVA) FOR ENERGY AND COMMODITY DERIVATIVES
Figure 15.1 Steps in the CVA process
Calculate Allocate
Determine Calculate portfolio
credit risk
CVA CVA for each CVA to
adjusted
method counterparty individual
curves
deals
Source: NQuantX LLC
Credit-adjusted rate curves

In order to calculate CVA, one of the critical set of inputs is the credit-
adjusted curves for the parties in the deal, which reflect the rates at
which each counterparty is able to borrow money for different matu-
rities in the capital markets.
There are several alternatives for creating credit-adjusted curves,
such as using credit default swap (CDS) spreads, corporate bond
yield spreads, and default probabilities from rating agencies, hybrid
models and internal rating systems. Table 15.1 provides a summary
of the alternative inputs.
Figure 15.2 Encana CDS spreads for various maturities (September

2007September 2012)
500
1Y 3Y 5Y 7Y 10Y
450
400
CDS Spreads (basis points)
350
300
250
200
150
100
50
0
08
09
11
7
12
01
00
20
20
20
20
/2
/2
9/
9/
9/
9/
09
09
/0
/0
/0
/0
/
/
03
03
03
03
03
03
391
Table 15.1 Main inputs used to build credit-adjusted rate curves
Benefits Cons
CDS spreads Research shows CDS premium Empirical research points that CDS
changes provide earlier warning spreads tend to overestimate
signals of credit risk problems. probability of default (PD).
Very liquid markets for some of the Not all counterparties have traded
larger firms. CDS.
Changes in credit spreads driven by
non-credit risk factors (eg, liquidity,
risk premium).
Bond yield Very liquid markets for some Less liquid than CDS.
spreads counterparties. Yield spreads changes driven by
Credit spreads can be derived from non-credit risk factors.
corporate bond yields.
Historical More stable than market based Not market-based.
default assessments. Slow to react to changing
probabilities Readily available for most conditions.
counterparties and sectors based on
external rating.
Hybrid More reactive than external ratings or Not directly based on credit market
models historical probabilities. assessments.
Likely to follow changes in CDS
and bond yields.
Internal Incorporates information from different More subjective elements.

ratings sources. Not necessarily market-based.
Consistent with internal credit Slow to react to changing
assessments. conditions.
Source: NQuantX LLC
CVA methods
There are several ways of calculating CVA, but we can group the
various methods in three categories. The first is the discount rate
adjustment, which requires the use of credit risk-adjusted discount
curves to calculate fair values. A common way of creating the credit
risk-adjusted curves for a given counterparty is by adding the CDS
spread to the risk-free rate curves used for present value calculations.
Figure 15.3 shows the CDS spreads for various North American
firms, as well as the zero-coupon risk-free curve and the risk-
adjusted curves. The risk-free rate curve is the US dollar zero coupon
curve for September 3, 2012. The credit-adjusted curves for each
entity in Figure 15.3 are created by adding the CDS spread to the
zero-coupon rate for each maturity.
392
Figure 15.3 Sample CDS spreads and credit risk-adjusted curves

9.000%
8.000%
7.000%
6.000%
5.000%
4.000%
3.000%
2.000%
1.000%
0.000%
0 2 4 6 8 10
Risk free Morgan Stanley
Chesapeake Encana
Source: NQuantX LLC
The discount rate adjustment is the most widely used method for
fair value reporting due to its simplicity. One of the main shortcom-
ings is that the credit exposure is assumed to be static, and therefore
the CVA is not dependent on the volatility of the mark-to-market
(MtM) of the deal, making it more likely to underestimate the credit
risk of the instrument.
The second type is known as the exponential CDS default method,
and requires the estimation of probabilities of default and recovery
rates. We can approximate the probability of default from quoted
CDS spreads for a given term by applying the following formula:
! CDSSpread t ( years) $
PD = 1! e # ! &
" (1! R ) 10000 %
where:
PD is the probability of default;

CDS Spread is the credit spread in basis points;
393
T (years) is the maturity of the CDS measured in years; and

R is the recovery rate (a common assumption made in published
CDS spreads is that the recovery rate is 40%).
Once the probabilities of default and recovery rates are estimated,

we can calculate the CVA for the deal based on the fair value using
the risk-free rate multiplied by the probability of default and the loss
given default (LGD), which is 1 less the recovery rate.
CVA = MtM(risk free) PD (1 R)
Again, the credit exposure is assumed to be static with this method,

so potential future exposures will tend to be underestimated.
The third method for calculating CVA is the exposure-based
approach, which requires an estimation of expected exposures (both
positive and negative) over the life of the deal, default probabilities
and recovery rates at different time steps. Exposure-based methods
combine market and credit risk elements.
In exposure-based methods, in addition to the forward curves and
other market variables, CVA is a function of the volatility of market
prices over time, the timing of cash inflows and outflows, the exis-
tence of collateral and netting agreements, the term structures of
default probabilities, as well as the expected recovery values.
Expected future positive and negative exposures can be calculated
using closed-form solutions or simulation methods.
Following Stein (2012), CVA is calculated as:
T
CVA = (1! R ) ! 0 S (t ) P (t ) dt
where S(t) is the expected exposure of the deal at time t, and P(t) is
the default probability time function.
As an alternative to calculating the full integral, we can divide the

time interval [0,T] into periods [ti, ti+1] and select ti [ti, ti+1].
A common choice for the time unit in each time interval is the
average between the two points:
T
T
CVA = (1! R ) ! 0 S (t) P (t )dt " (1# R ) ! S ( ti )P (ti )
t=0
where
ti+1
P (ti ) = ! ti
P (t ) dt
394
is the probability of default in the interval [ti, ti+1].

If we simplify to just one period, CVA and DVA for a one-period
horizon can be calculated applying the following formula:
CVA = PDcpty EPE LGDcpty Spreadcpty EPE
DVA = PDown ENE LGDown Spreadown ENE
where PD is probability of default, EPE and ENE are the expected

positive and negative exposures, respectively.
We can see the expected positive exposure (EPE) and the expected
negative exposure (ENE) profile for a set of nettable contracts with a
given counterparty in Figure 15.4. The EPE is used for CVA calcula-
tions and represents the expected exposure in the event of default by
our counterparty at different horizons. The ENE represents the
expected exposure in the event of our own firm defaulting.
Exposure-based methods are the more comprehensive ones, as
they overcome the shortcomings of assuming that the credit expo-
sure is static. The three main methods are summarised in Table 15.2.
Figure 15.4 Expected positive exposure and expected negative exposure profiles
$30,000,000
$20,000,000
$10,000,000
Expected positive exposure (EPE)
$0
-$10,000,000
Expected negative exposure (ENE)
-$20,000,000
-$30,000,000
13
13
13
13
13
13
13
13
12
12
01
01
20
20
20
20
20
20
20
20
20
20
20
/2
2
1/
2/
3/
4/
5/
7/
8/
9/
0/
1/
1/
06
12
/0
/0
/0
/0
/0
/0
/0
/0
/1
/1
/1
/
/
01
01
01
01
01
01
01
01
01
01
01
01
01
Expected positive exposure Expected negative exposure
Source: NQuantX LLC
395
Table 15.2 Main CVA methods
CVA type Description

Discount rate After calculating the MtM, the value is discounted using the
adjustment (I) credit spread to calculate the CVA.
Discount rate The MtM is calculated discounting each of the flows using the
adjustment (II) risk-adjusted curves of one or two of the parties in the deal.
Exponential CDS Formula-based approach that introduces the CDS spread,
default method survival rates and recovery rates.
Exposure-based Involve the estimation of expected potential exposures over
methods different time horizons, probability of default and recovery
rates.
Source: NQuantX LLC
CVA at the counterparty level

To perform the credit adjustment to the value of a derivatives book,
we need to calculate the CVA and DVA for each counterparty in the
book. The net adjustment is the difference between the sum of the
CVA minus the sum of the DVA for all counterparties.
The calculation of the CVA and DVA for each counterparty is not
simply the sum of the individual trades CVA and DVA. This is
because, in order to calculate CVA at a portfolio level, it is necessary
to take into account netting, collateral and other credit risk mitigants.
We can use any of the CVA methods to perform calculations at the
portfolio level. The main difference is that, instead of just adding the
individual MtM and potential exposures of each instrument, we
need to perform those calculations after taking into account any
credit mitigants. For example, if we use the discount rate method, we
could calculate the net exposure for each counterparty after applying
netting and collateral rules, and then calculate CVA by multiplying
the net exposure times the credit spread.
For portfolio level calculations, it is common to perform simula-
tion of risk exposures at the counterparty level for multiple scenarios
after taking netting and collateral into account. The steps involved in
calculating portfolio CVA in a simulation framework are shown in
Figure 15.5.
Although CVA calculations are based on EPE and ENE, credit risk
charges are often based on potential future exposures at a high confi-
dence level. Figure 15.6 shows a potential future exposure (PFE)
396
Table 15.3 CVA report at the counterparty level
Counterparty MtM # trades Collateral Net exposure CVA % MtM
Citigroup US$5,034,352 US$45 US$2,500,000 US$2,534,352 US$74,375 1.48%

BNP Paribas US$3,775,764 US$32 US$1,000,000 US$2,775,764 US$74,171 1.96%
Goldman Sachs US$4,14,604 US$12 US$414,605 US$11,127 2.68%
Glencore US$100,651 US$7 US$100,651 US$2,753 2.73%
JP Morgan US$9,165 US$23 US$9,165 US$258 2.81%
Shell US$6,873 US$11 US$6,874 US$158 2.30%
Source: NQuantX LLC

397
Figure 15.5 Steps to calculate CVA in a simulation framework
1. Generate spot and forward curve scenarios for multiple time steps
2. Value each deals for each scenario time step
3. Apply netting rules at the counterparty level for each scenario time step
4. Calculate net exposure for each scenario time step
5. Repeat process for multiple scenarios
6. Calculate exposure profile metrics (EPE, PFE, )
7. Calculate credit valuation adjustment at the counterparty level
Source: NQuantX LLC
report at the counterparty level with the PFE profiles for each coun-
terparty.
From portfolio CVA to deal CVA

When two entities enter into a series of transactions, they also
exchange an implicit option to default. CVA is the price or cost of
credit risk for a deal or portfolio with a given counterparty.
CVA can be calculated at the individual transaction or at each
counterparty portfolio level. The overall CVA for the exposures with
a given counterparty is not simply the sum of the individual deal
CVA, because of the need to take into account credit risk mitigation
rules that apply to those exposures. For example, a trading entity
may have several deals with the same counterparty that have large
stand-alone CVAs, but if those exposures offset each other and there
are netting agreements in place, the overall portfolio CVA will be
considerably lower than the sum of the individual CVAs.
Despite the non-additive nature of portfolio CVA, it is possible to
398
Figure 15.6 Potential future exposure (PFE) at the counterparty level

US$16,000,000
US$14,000,000
US$12,000,000
US$10,000,000
US$8,000,000
US$6,000,000
US$4,000,000
US$2,000,000
US$
1M 3M 6M 12M 2Y 3Y
Citigroup Goldman Sachs JP Morgan
BNP Paribas Glencore Shell
Source: NQuantX LLC

399
allocate it among the various portfolio constituents in a similar

fashion than we can calculate marginal VaR for individual portfolio
components to determine the contribution to diversified VaR of a
given trade or strategy. The allocation of the portfolio CVA into each
individual trade is important for individual hedge accounting desig-
nations, as well as improving the understanding of the marginal
impact of individual trades on the overall CVA.
The process to allocate portfolio CVA into individual components
consists of determining the marginal contribution of each trade to the
portfolio CVA. The marginal contribution of a given trade could be
positive, negative or neutral, depending on the change in the port-
folio CVA before and after including that trade.
CVA allocation approaches

There are various approaches to allocating CVA among individual
portfolio constituents. It is important to understand the differences
between them because the choice of allocation method may have
material implications on how credit risk adjustments of assets and
liabilities are reported, as well as hedge effectiveness tests and quali-
tative derivatives disclosures.
The most common CVA allocation methods are:1
In-exchange or full credit: the stand-alone CVA for each deriva-

tive instrument is directly applied. There is no need to calculate
portfolio CVA or apply any allocation methodologies.
Relative fair value: the portfolio CVA is allocated to each deriva-
tive instrument according to the sign and magnitude of the fair
value of each derivative asset and liability.
Relative credit adjustment: the portfolio CVA is allocated to each
derivative instrument according to the sign and magnitude of
the stand-alone CVA for each derivative asset and liability.
Marginal contribution: the portfolio CVA is allocated to each
derivative instrument based on the sign and magnitude of the
marginal contribution to CVA of each individual derivative asset
and liability.
A numerical example of the application of the relative credit-

adjustment method is shown in Tables 15.4, 15.5 and 15.6; note that
the application of this method in particular is investigated because
400
Table 15.4 Stand-alone CVA report
Counterparty BP Trading
Mark-to- Stand-alone Relative Credit-adjusted

Stand-alone CVA market CVA weights MtM
WTI swap US$1,243,550 US$10,724 109.63% US$1,232,826

Brent swap US$672,946 US$5,114 52.27% US$667,832
Natural gas swap US$(831,503) US$(6,363) 65.04% US$(825,141)
Natural gas basis swap US$97,690 US$302 3.08% US$97,388
Totals US$1,182,682 US$9,777 100% US$1,172,905

Portfolio Undiversified Credit-adjusted
MtM CVA MtM
Source: NQuantX LLC
Table 15.5 Portfolio-level CVA
Portfolio Netting ALL
Collateral held/(posted) US$1,000,000
Counterparty Mark-to-market Net exposure Portfolio CVA Credit Adjusted MtM
BP Trading US$1,182,682 US$182,682 US$3,654 US$1,179,029
Source: NQuantX LLC
the relative credit adjustment method is relatively easy to imple-

ment, and superior to the in-exchange or full credit, and the relative
fair value methods.
The first step involves calculating the stand-alone CVA for each
individual trade. If we add the stand-alone CVAs at the trade level,
we can calculate the undiversified portfolio CVA assuming that
the portfolio consists of non-nettable, non-collateralised trades, effec-
tively ignoring credit-mitigation effects. The magnitude and sign of
each stand-alone CVA determines the relative weights for the deals
that are part of that counterpartys portfolio. Table 15.5 shows the
deal-level breakdown of MtM and CVA of the portfolio with trades
made with EDF Trading as the counterparty. The bottom row shows
the portfolio MtM, the undiversified CVA as a sum of stand-alone
CVA, as well as the credit-adjusted MtM. The relative percentage
weights are calculated by dividing each individual CVA by the undi-
versified portfolio CVA.
401
Table 15.6 Allocation of portfolio CVA to individual trades
Trade # Mark-to-market CVA Marginal CVA Credit-adjusted

weights MtM
WTI swap US$1,243,550 109.63% US$4,005 US$1,239,545
Brent swap US$672,946 52.27% US$1,910 US$671,036
Natural gas swap US$(831,503) 65.04% US$(2,376) US$(829,127)
Natural gas basis US$97,690 3.08% US$113 US$97,577
swap
Portfolio MtM Portfolio CVA Portfolio credit-
(risk free) bilateral adjusted MtM
US$1,182,682 100.00% US$3,652 US$1,179,031
Source: NQuantX LLC
The next step consists of calculating the portfolio CVA based on

the unsecured exposure with each counterparty, which is estimated
after taking into account any relevant netting and credit-mitigation
tools. Portfolio CVA calculations can be made using current or
expected exposure methods for the various trades in the portfolio.
The combined MtM exposure, the net exposure and the portfolio
CVA are shown in Table 15.5. In our example, as the current expo-
sure is largely collateralised, the portfolio CVA is substantially lower
than the sum of the stand-alone CVAs. The last column shows the
combined credit-adjusted MtM for the exposures with BP Trading.
As a final step, we can calculate the portfolio CVA portion that
will be allocated to each individual deal by multiplying the relative
weights calculated in Table 15.4 times the portfolio CVA. This is
shown in Table 15.6.
A comparative analysis of the different CVA allocation method-
ologies is shown in Table 15.7. The relative credit adjustment and the
marginal contribution approaches are the most accurate ones, but
require the calculation of new metrics such as stand-alone and
marginal CVAs. Under those approaches, the CVA of a portfolio of
trades with a given counterparty is allocated among its individual
constituents according to the relative weight of each individual
stand-alone or marginal CVA.
The state of the practice of CVA reporting by energy trading firms
is still in its infancy. Several publicly traded entities still do not calcu-
late portfolio CVA and most do not allocate portfolio CVA into
402
individual deals. The main reason given is that the allocation of

credit risk would not materially impact the fair value or the determi-
nation of the hedge effectiveness of a hedging relationship.
Credit limits, charges and CVA desks

Traditionally, credit risk management at energy and commodity
trading firms has been a passive exercise, where credit risk managers
set and monitor exposure limits and traders can continue dealing
with a given counterparty as long as those limits are not breached.
Those limits are often defined based on the current exposure at the
counterparty level, which is generally calculated by taking the mark-
to-market value of open positions, adding account receivables and
accounts payable, and open settlements and applying appropriate
netting rules and credit risk mitigation tools. In addition, the limits
may be scaled as a function of the evolution of the creditworthiness
of the counterparty. A sample credit report is shown in Table 15.8
with limits based on current exposures as a function of the counter-
partys internal rating.
One of the main problems of this binary approach based on
Table 15.7 Portfolio CVA allocation methodologies
Approach Advantages Limitations

In-exchange or Simplicity of approach. No Netting and collateral agreements
full credit need to perform portfolio not taken into account.
CVA calculations. Overestimates credit risk at the
individual instrument level.
Relative fair Simple to implement. Only Allocation weights based on fair
value requires ability to calculate value may differ considerably
CVA at the counterparty from those based on stand-alone
portfolio level. or marginal CVA.
Relative credit Relatively simple to Stand-alone CVA may not
adjustment implement. Allocation based represent marginal contribution,
on actual CVA of each but often a good proxy.
instrument.
Marginal Technically, the most Marginal CVA contributions may
contribution accurate method. Allocation be volatile. Requires ability to
based on marginal impact on calculate marginal CVA.
portfolio CVA of each
derivative instrument.
Source: NQuantX LLC
403
Table 15.8 Sample credit risk limit report (US$ 000s)
Counterparty Internal Current Limit Limit Potential PFE/

rating exposure usage future limit
exposure
Barclays 3 US$2,092 US$5,000 42% US$7,540 151%

Morgan Stanley 4 US$4,021 US$3,000 134% US$5,650 188%
J Aron & Co 3 US$(461) US$5,000 0% US$12,350 247%
Shell 2 US$(18) US$2,000 0% US$325 16%
BP 3 US$1,682 US$5,000 34% US$4,500 90%
Exxon 1 US$1,729 US$10,000 17% US$3,450 35%
Southern Company 5 US$714 US$1,000 71% US$1,250 125%
limits is that it creates a situation where traders become the frontline

credit risk managers for the firm while their goals become that of
maximising profitability, not managing that risk. As a result, the firm
may end up with excessive credit risk concentrations to a given
industry, geographical region or tenor due to the combined trader
exposures. These concentrations could be exacerbated in the future if
the exposures are driven by the same market risk factors.
To complement static current exposure limits, many credit risk
departments have also added, as part of the limit structure, forward-
looking exposure and risk metrics such as maximum PFE at a given
confidence interval. Those metrics can alert credit risk managers of
the exposures that may grow beyond the maximum thresholds as a
result of market fluctuations. In the last column of Table 15.8, we can
see that the existing exposures with Barclays, Morgan Stanley, J.
Aron and Southern Company may exceed the credit limits in the
future, even although they are within the approved current exposure
limits.
The next level in sophistication of the credit risk function involves
designing a system of credit risk charges that can ensure that risk
takers become actively involved in the risk management process. A
system of credit risk charges and reserves can assist the front office in
setting the right price for each new transaction based on the risks
incurred by the firm, and also to create a fund with reserves to
protect the firm against future credit losses.
There are two main types of counterparty risk charge systems. The
first is based on setting upfront charges for each new deal as a func-
tion of the potential credit losses over the life of the deal using CVA
404
Table 15.9 Sample trader-level allocation of pay-as-you-go charges for a single

counterparty
Counterparty Current Credit spread Credit Daily

exposure (bp) spread charge
Credit Suisse 2,091,546 245 2.45% US$140.39
Trader MtM Charge Daily

allocation charge
M. Smith US$1,235,000 36.5% US$51.31

J. Arnold US$(212,456) 0.0% US$
C. White US$324,500 9.6% US$13.48
M. Ford US$750,700 22.2% US$31.19
S. Chance US$1,069,002 31.6% US$44.41
Totals US$3,379,202 100.0% US$140.39
Source: NQuantX LLC
or PFE metrics.2 Traders often strongly oppose these types of

systems due to the fact that charges for the full life of the deal are
applied upfront, and if the deal is reversed before maturity that
would result in overcharges.
The other type is known as pay-as-you-go, and consists of
applying a daily credit risk charge based on current unsecured credit
exposures.3 Pay-as-you-go systems are relatively easy to implement
as the only metrics required are the unsecured credit exposure
against each counterparty and the daily cost of capital of that
counterparty.
Pay-as-you-go systems are conceptually appealing for many
trading organisations, but the success of the implementation of the
credit risk charge system is often dependent on the details. For
example, if traders are charged on the stand-alone credit risk of each
deal, they are likely to be overcharged on an overall basis and there-
fore be at a competitive disadvantage over other firms.
The best systems strike the right balance in terms of taking into
account diversification benefits at the counterparty portfolio level
and also impact the behaviour of the risk takers directly. Although
pay-as-you-go systems apply daily charges, the credit risk managers
can provide traders with the distribution of their maximum potential
future credit charges before they conduct a new deal group using
forward-looking probability-based credit metrics such as PFE.
405
Active risk management with CVA desks

As a response to the increased magnitude and complexity of the
credit risk dimension in derivatives trading, many firms have
created internal groups, known as CVA desks, with a mandate to
measure, price and manage counterparty risks. CVA desks act as
specialised units that manage the credit risk of the consolidated
exposures with each major counterparty.
Some CVA desks are in charge of implementing the system of
credit charges and reserves for the firm. A sample decision flow
diagram is shown in Figure 15.7. When a trader wants to conduct a
new deal, the CVA desk determines the cost to protect the credit risk
of that new deal and communicates it to the trader. The credit charge
can be calculated based on the size, liquidity and duration of the
exposure, the credit risk mitigation tools in place (netting, collateral,
guarantees, etc), as well as the creditworthiness of the counterparty.
If the trader decides to go ahead with the transaction, it may pass the
fees to the counterparties by pricing the deal taking into account the
cost of protection.
The CVA desk can also play an active role by encouraging traders
to use collateral to minimise credit exposures, and also to trade with
certain counterparties that reduce overall portfolio exposures. In
order to do so, they need to have the ability to calculate credit risk on
Figure 15.7 Active credit risk management with CVA desks
Trader approaches CVA desk before conducting new deal
CVA desk calculates fee for credit risk protection
Trader pays CVA premium

CVA
Trader
CVA collects fee and desk
protects credit
exposure
CVA purchases
Credit protection for

unsecured exposures
from trading position
markets
Source: NQuantX LLC
406
an aggregate portfolio from multiple potential trades, as well as to

determine credit charges, pricing, hedging and reserves from those
hypothetical trades. In certain instances, the CVA desk could poten-
tially allow certain deals to be priced at a discount if they reduced the
counterparty risk from a portfolio perspective.
In some financial institutions, the CVA desk is structured as a
profit centre whose traders attempt to take active credit bets. For
most energy and commodity firms, however, the starting goal of the
CVA desk is likely to be the active management of credit exposures
with an emphasis on credit risk hedging and loss minimisation.
CONCLUSION
Counterparty and liquidity risk management in energy and
commodity portfolios is undergoing a revolution driven by efforts to
price and hedge those risks. CVA is gradually becoming an integral
part of the risk management process of energy and commodity
trading firms by helping to adjust valuations, and also in setting
credit and liquidity risk charges.
There are various methods to calculate CVA, from relatively
simple discounting of cashflows using credit risk-adjusted curves to
the more complex simulation-based potential exposures coupled
with default and recovery rates.
Valuation and risk measurement models and systems that fail to
incorporate counterparty and liquidity risk adjustments are ignoring
a material risk that could impact the accuracy of fair value and P&L
calculations, as well as the validity of risk metrics used throughout
the firm.
The implementation of CVA at the valuation and risk measure-
ment level introduces other risks, such as potentially higher P&L
volatility and increased model risk arising from potential CVA esti-
mation errors, but the benefits often outweigh the risks. Firms with
the ability to price and manage credit risk proactively will have a
competitive advantage over those that continue to manage those
risks in a passive and reactive fashion.
1 For a more detailed explanation of the different methods, we recommend:

PricewaterhouseCoopers, 2008, Consideration of Credit Risk in Fair Value Measurements:
An Addendum to PwCs Guide to Fair Value Measurements.
2 The first part of this chapter covers the main CVA calculation methodologies.
3 For more details, see Humphreys and Shimko (2005).
407
REFERENCES
Blanco, C., 2010, Collateral, Cash Flow & Earnings at Risk: Time to update your risk
metrics and policies?, Commodities Now, July.
Blanco, C., 2010, Credit Valuation Adjustment for Commodity Derivatives, Commodities
Now, December.
Blanco, C., K. Dowd, R. Mark and W. Murdoch, 2006, Credit Risk Measurement and
Management for Energy Firms, in P. C. Fusaro (Ed), Professional Risk Managers Guide to
Energy and Environmental Markets (New York, NY: McGraw-Hill): pp 6982.
Humphreys, B. and D. Shimko, 2005, Pay As You Go, Energy Risk, January.
Stein, H., 2012, Counterparty Risk, CVA, and Basel III, Columbia University Financial
Engineering Practitioners Seminar, March.
408
16
The Past, Present and Future of
Chinas Futures Market: Trading
Volume Analysis
Wang Xueqin
Zhengzhou Commodity Exchange
This chapter will review the development of Chinas futures market

since the early 1990s, with corresponding analysis of futures trading
volume. It will discuss the two phases of clean-up and rectification
necessitated by the development of Chinas futures markets and the
establishment of its regulatory regime in the 1990s. In this context, it
analyses the trading volumes of Chinas futures market between
2000 and 2011, particularly trading characteristics in 2011. This
analysis will help to explain why the trading volume of futures and
options rose worldwide but dropped in China during the same
period, as well as the reason why Chinas futures market has latterly
been experiencing greater success. The last part of this chapter will
offer some predictions based on the huge potential for furthering the
futures market in China. These predictions include a trend towards
loosening product control, incorporating exchanges and the further
opening up of the futures market in China. Finally, this chapter will
conclude that a futures market with a stable trading volume will
conform to the principle of making progress while ensuring
stability. Such progress will be aided by research on the insurance
function of futures options, grasping the developing direction of
options early on and accelerating the innovation of futures products
and new futures business.
409
THE PAST
In 1990, with the approval of the State Council, the Zhengzhou Grain
Wholesale Market1 was founded as the first futures pilot programme
in China. Since then, Chinas futures market has continued to
develop, moving from chaos to governance including two
phases of clean-up and rectifications in the 1990s, as well as stan-
dardisation development in the 2000s. The futures market has grown
steadily in scale, with laws and regulations increasingly imple-
mented, market functions well-performed and international
influence becoming even more significant. This has been especially
true since 2004, with more futures contracts being listed and traded,
forming an almost complete commodity futures market system
consisting of agricultural, metal, energy and chemical products.
The first 10 years (19902000)

In the 1990s, along with the establishment of a market-based
economic policy, China gradually opened its free commodity
market, resulting in frequent fluctuations in commodity prices. The
government began to develop a wholesale market of food, metals,
materials and other commodities, and futures trading of these prod-
ucts was gradually promoted. For a time, the commodity wholesale
markets and futures exchanges flourished everywhere in China.
From 1990 to 1993, the number of futures exchanges increased
rapidly, while futures contracts became multiply listed and the
markets were over-speculated. There were then more than 50 futures
exchanges in China, leading to vicious competition between the
exchanges. Disorderly market trading, market manipulation and
insider trading also occurred frequently during that time.2 Over a
thousand futures commission companies were founded, although
their operation and management verged on the chaotic. Customers
were often cheated and the interests of investors violated. The
brokerage services for overseas futures trading developed too fast to
be brought under regulation in a timely manner. The futures market
entered a stage of blind development.
The first phase of clean-up and rectification

On November 4, 1993, the State Council of the Peoples Republic of
China issued the Notice to Resolutely Stop the Blind Development
of the Futures Markets, and started the first clean-up and rectifica-
410
Figure 16.1 Estimated Chinese trading volume (US$ trillion US$ equivalents) and number of listed futures
THE PAST, PRESENT AND FUTURE OF CHINAS FUTURES MARKET: TRADING VOLUME ANALYSIS
30 60
Trading turnover
Trade volume US$ trillion US$ equivalents Listed Contracts (right axis)
25 50
20 40
Number of listed futures

15 30
10 20
5 10
0 0
3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2
199 199 199 199 199 199 199 200 200 200 200 200 200 200 200 200 200 201 201 201
411
tion in the futures market. The key measures adopted included

reducing the number of futures exchanges and futures products, as
well as suspending bulk trading for several commodities that were
closely related to the peoples livelihood but the price of which were
still subject to policy control such as steel, sugar, coal, petroleum
and vegetable oil. On May 18, 1995, the Treasury bond futures
contract was also suspended. By the end of the first clean-up and
rectification, only 14 exchanges were approved to remain in busi-
ness, while just 35 futures contracts were approved, divided into
formally listed contracts and trial-based contracts. Figure 16.1 shows
the number of listed futures and total futures trading notional
turnover by year.
Therefore, the main measures taken in the first clean-up and recti-
fication were:
reduce the amount of futures exchanges and futures contracts;

suspend bulk trading for several commodities that were closely
related to the peoples livelihood and the prices and were still
subject to policy control, such as steel, sugar, coal, petroleum and
vegetable oil;
suspend trading for Treasury futures on May 18, 1995;
suspend approval for futures companies, review futures compa-
nies and implement a licence system;
strictly control overseas futures trading;
strictly control the participation of state-owned enterprises and
institutions, and financial institutions in futures trading; and
establish a centralised and unified regulatory regime.
The second phase of clean-up and rectification

After the first phase of clean-up and rectification, there were still too
many futures exchanges and futures brokerage firms with problem-
atic operations, and a few institutions and individuals were
manipulating the market to make exorbitant profits. There were
numerous incidences of illegal futures trading in overseas markets.
The supervisory administration had weak supervision powers and
retrograde supervisory methods.
In August 1998, the State Council issued the Circular on Further
Rectification and Regulation of Futures Market, in which the princi-
ples of continuing the pilot project, strengthening regulation,
412
Table 16.1 Exchanges, brokerage firms, listed commodities and turnover in Chinese
futures
Year Futures Brokerage Listed Trading Trading

exchanges firms commodities turnover volume
contracts (in US$100 million)* (contracts)
1990 1 - - - -
1991 2+ - - - -
1992 5+ - - - -
1993 32+ 1000** 52+** 443.18 4,453,450
1994 14 144 35 2536.23 60,553,600
1995 14 330 35 8071.05 318,060,350
1996 14 329 35 6751.14 171,283,850
1997 14 294 35 4909.36 79,381,600
1998 14 278 35 2966.87 52,227,850
1999 3 213 12 1793.18 36,819,550
2000 3 178 12 1290.71 27,305,350
2001 3 184**** 12 2419.34 60,231,750
2002 3 187 11*** 3169.35 69,716,316
2003 3 176 11 8698.96 139,932,111
2004 3 172 11 11792.56 152,848,800
2005 3 166 11 10790.40 161,423,761
2006 3 164 14 16857.65 224,737,051
2007 3 165 18 32883.02 364,213,397
08 3 163 19 57716.05 681,943,551
09 3 164 23 104743.76 1,078,714,909
10 4 163 24 248087.05 1,566,764,672
11 4 161 27 220727.81 1,054,088,664
12 4 161 31 274675.97 1,450,462,383
Source: Compiled from data in CSRC, 2001, China Securities and Futures Statistical Book,
apart from the number of exchanges in 1993 being taken from 100 Questions and Answers of
Futures Operation (China Material Publisher): p.138 (while the number of exchanges is put at
32 in this report, it was generally recognised that there were over 40 exchanges operating in
1993; the number of exchanges in 1990, 1991 and 1992, and the number of brokerage firms
in 1993, are the authors estimates).
*Trading turnover in Chinese yuan (Yn) was converted to US dollars at an exchange rate of
6.23 Yn/US$.
**The 19932001 listed commodities contracts data and the number of 19932000 brokerage
firms was taken from Xueqin and Gorham (2002).
***200212 listed commodities contracts dates came from:
http://www.cfachina.org/news.php?classid=108.
****The 2001 brokerage firms number was taken from:
http://www.yafco.com/show.php?contentid=40670, while the 200212 brokerage firms
number came from: China Futures Industry Development Report and the CSRC.
standardising according to law and preventing risk were determined

for the second clean-up and rectification in Chinas futures market.
The main measures included the further rectification, cancellation
and consolidation of futures exchanges. Three futures exchanges
413
were retained, in Shanghai (Shanghai Futures Exchange, SHFE),

Zhengzhou (Zhengzhou Commodity Exchange, ZCE) and Dalian
(Dalian Commodity Exchange, DCE), and the exchange manage-
ment system was upgraded. In order to fully perform the functions
of price discovery and hedging for futures markets and further
contain excessive speculation, the listed futures contracts were
reduced from 35 to 12, with 23 futures contracts de-listed. The 12
commodities remaining were copper, aluminium, soybeans, wheat,
soybean meal, mung bean, natural rubber, plywood, long-shaped
rice, beer barley, red bean and peanuts.
The measures for the second phase also included the level of
minimum registered capital for futures brokerage firms being
increased to US$4.82 million. In 1998, after the increase of registered
capital, there were about 180 futures brokerage firms remaining in
normal operation. In addition, the China Securities Regulatory
Committee (CSRC) was created and Interim Regulations on
Administration of Futures Trading implemented.3
Determine the regulatory regime

After the clean-up and rectification, CSRC was appointed as the
centralised and unified regulatory commission for Chinas futures
market, a regulatory change that created steady and well-
functioning futures markets. In addition, Chinas futures market
withstood the impact of the financial crisis of 200708; due to the
influence of the crisis, price volatility increased, especially in
commodity futures. The global financial crisis also impacted upon
the growth rate on Chinas futures market. In 2009, this plunged to
20.72% from 87.24% in 2008, down 66.52%, showing that the
centralised and unified regulation had been successful and deserved
high credit, and also that it provides useful experience for the devel-
opment and supervision of other commodity futures and financial
derivatives in China.
On November 24, 1998, the CSRC approved six modified contracts
for soybean, wheat, mung bean,4 copper, aluminium and natural
rubber, and on November 27, it approved modified contracts for
wheat and mung bean, to be re-listed and traded again. These
approvals marked a turning point in Chinas futures markets in the
latter stages of its first decade. The second significant event was the
implementation of Interim Regulations on Administration of
414
Figure 16.2 Regulations on administration of futures trading
CSRC
36 CSRC were distributed to various places in 1998
China Futures Association China Futures Margin

was founded in 2000 Monitoring Centre in 2008
Shanghai Futures Dalian Commodity Zhengzhou China Financial

Exchange Exchange Commodity Exchange Futures Exchange
Futures Trading,5 and the third was the founding of the China
Futures Association on December 28, 2000.6
The second 10 years (200010)

In the first ten years of 21st century, Chinas futures market devel-
oped rapidly. The most significant events included new futures
contracts listed for trading, futures companies gradually opening up
and financial futures being launched successfully.
Figure 16.3 Chinese yearly futures trading volumes (millions of contracts)

1600
Trading volume (millions of contracts)
1400
1200
1000
800
600
400
200
0
1993 1995 1997 1999 2001 2003 2005 2007 2009 2011
415
Table 16.2 Futures contracts list for trading on China futures exchanges
Exchanges Contracts Time Trading volume Trading volume

in (2011) (2012)
Shanghai Aluminium 1992 9,953,918 3,942,680

Futures Copper March 1993 48,961,130 57,284,835
Exchange Natural rubber November 1993 104,286,39 75,176,266
Fuel oil August 25, 2004 1,971,141 9,132
Zinc March 26, 2007 53,663,483 21,100,924
Gold January 9, 2008 7,221,758 5,916,745
Deformed steel bar March 27, 2009 81,884,789 180,562,480
Wired rod March 27, 2009 3,242 2,717
Lead March 24, 2011 293,280 68,646
Silver May 10, 2012 21,264,954
Volume 308,239,140 365,329,379
Zhengzhou Excellent strong gluten

Commodity wheat March 28, 2003 7,909,755 25,802,102
Exchange No. 1 cotton June 1, 2004 139,044,152 21,016,438
Sugar January 6, 2006 128,193,356 148,278,025
PTA (pure terephthalic acid) December 18, 2006 120,528,824 121,245,610
Rape oil June 8, 2007 4,320,115 6,248,568
Regular white wheat March 24, 2008 152,901 6,262
(hardwhite
wheat)
Early long-grain non
glutinous rice April 20, 2009 5,925,454 3,838,320
Methanol October 28, 2011 316,107 3,797,412
Glass December 3, 2012 - 16,136,920
Rape seed December 28, 2012 - 137,084
Rape meal December 28, 2012 - 421,207
Volume 406,390,664 347,028,203
Dalian Commodity Soybean meal July 17, 2000 50,170,334 325,876,653

Exchange No. 1 yellow soybean March 15, 2002 25,239,532 45,475,425
Corn September 22, 2004 26,849,738 37,824,356
No. 2 yellow soybean December 22, 2004 10,662 10,400
Soybean oil January 9, 2006 58,012,550 68,858,554
LLDPE (linear low-density
polyethylene) July 31, 2007 95,219,058 71,871,537
Palm oil October 29, 2007 22,593,961 43,310,013
PVC (polyvinylchloride) May 25, 2009 9,438,431 6,900,153
Coking coal April 15, 2011 1,512,734 32,915,885
Volume 289,047,000 633,042,976
China Financial ShanghaiShenzhen 300

Futures Exchange stock index Apr. 16, 2010 50,411,860 105,061,825
Volume 50,411,860 105,061,825
Futures trading
volume in China Total 1,054,088,664 1,450,462,383
New futures contracts promoted

By the end of 2011, there were 31 futures contracts listed in four
futures exchanges in China, covering agricultural, industrial, energy
and financial products. Futures contracts for Treasury bonds, crude
oil, coking coal and even potato and egg were also being considered.
416
In addition, steam coal futures contracts at the Zhengzhou

Commodity Exchange, coking coal and iron ore futures contracts at
the Dalian Commodity Exchange, crude oil futures contracts at the
Shanghai Futures Exchange and debt futures contracts at the China
Financial Futures Exchange were prepared and subject to regulatory
approval (see Table 16.2).
Futures companies gradually opened up

Chinese brokerage companies also became qualified for overseas
futures business. At the end of June 2001, the CSRC officially
published Administrative Method for Overseas Futures Hedging
Business of State-owned Enterprises, a list of state-owned enter-
prises that were allowed to participate in overseas futures hedging
business. By end-2005, 31 enterprises obtained the qualification of
overseas futures business.
Foreign capital participated shares into Chinese futures companies

By 2013, there were only three Sinoforeign joint venture futures
companies in China: Galaxy Futures, CITIC Newedge and JPMorgan
Futures. On December 2, 2005, ABN of Royal Bank of Scotland was
approved to set up the first joint venture futures company with
Galaxy Futures in China. The foreign party, ABN Financial Futures
Asia Co Ltd, held 40% of the shares, while on the Chinese side,
Galaxy Securities held the remaining 60% of shares. CITIC Newedge
was formerly known as CITIC Futures in January 2007, overseas
strategic partner Newedge Group was launched, founded by Socit
Gnrale and Banque de Llndochine with a 5050% ownership ratio.
The futures business of the group was carried out by Banque de
Llndochine and the Pegasus Group. On September 26, 2007,
Guangzhou Zhongshan Futures was renamed as JPMorgan Futures,
with JPMorgan holding about 49% of shares indirectly.7
Chinese FCMs started to set up branches in Hong Kong

The China Securities Regulatory Commission approved six Chinese
future commission merchants (FCMs) to set up branches in Hong
Kong in March 2006, namely China International Futures, Yongan
Futures, GF Futures, Nanhua Futures, Jinrui Futures and Green
Futures.
417
Table 16.3 Latest trading volume of stock index futures in China financial
futures exchange
Year 2010 2011 2012 Total
CSI 300 91,746,590 50,411,860 105,061,825 247,220,275
Financial futures successfully launched

China Financial Futures Exchange (CFFEX), approved by the State
Council and the CSRC, was set up by five shareholders: Shanghai
Futures Exchange, Zhengzhou Commodity Exchange, Dalian
Commodity Exchange, Shanghai Securities Exchange and Shenzhen
Securities Exchange, with each of the shareholders contributing
US$16.05 million. The CFFEX was founded on September 8, 2006, in
Shanghai. On April 16, 2010, CSI300 stock index futures started to be
listed and traded on CFFEX.
Trading volume of Chinas futures market (200010)

Since the early 2000s, the trading volume of Chinese futures has
increased exponentially. The annual average growth rate of Chinas
futures market between 2000 and 2010 was 43.16%.
In 2001, the trading volume of Chinese futures stood at 60.23
Figure 16.4 Growth in trading (log scale) for Chinas futures market
100
Trade volume (US$ trillion, US$ equivalent)
10
0.1
0.01
3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2
199 199 199 199 199 199 199 200 200 200 200 200 200 200 200 200 200 201 201 201
418
million contracts. The trading volume of global futures and options

was 4.4 billion contracts, and trading volume of global futures was
about 1.8 billion. The trading volume of global commodity futures
and options was about 400 million, and the volume of global
commodity futures was about 380 million contracts.
A decade later, the trading volume of Chinese futures had grown
to 1.52 billion lots and global futures was about 11.2 billion. The
growth in size of Chinas markets as a percentage of the total global
market for all futures and for commodity futures is shown in Figure
16.5.
Global futures volumes in Chinas futures markets increased from
3.34% in 2001 to 13.6% in 2010. For the first half of that period, the
market share of China was no more than 5%. Over the period 2001
10, the ratio for futures and options increased from 1.37% to 6.82%,
illustrating the slower growth of options trading relative to futures
trading in China. During the 11th five-year plan, the market share of
Chinas futures market to global futures and options market
increased exponentially.
For commodity futures, the ratio of Chinas increased from 15.99%
of the global volume to 53.65% in 2010. When including options
volumes, Chinas market increased from 14.61% to 50.95% of global
futures and options volume in 2010.
To summarise, between 2001 and 2010 the rapid increase in
trading volume made Chinas futures market the largest commodity
Figure 16.5 China markets as a percentage of global trading volume

60
Commodity futures
50
Commodity futures and options
Futures
40
Futures and options
30
20
10
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
419
market in the world. On the other hand, the average annual share of
the Chinese futures market share to global futures and options
remained low, reflecting that the trading volume of Chinas futures
market in commodity options, stock futures and stock options are
not yet fully developed, with some products being totally absent
from Chinas derivatives market. In 2010, Chinas futures market
was one of the largest commodity futures market in the world,
accounting for a trading volume of 53.65% of the global commodity
futures market.
THE PRESENT
The micro characteristics of the market operation
The healthy development of the Chinese futures market has been
mainly based on the powerful spot market and an upward macro-
economic environment. Chinas consumption volume of nearly all
the commodities underlying the futures contracts ranks among the
global top three consuming nations (see Table 16.4).
Open interest
As research shows,8 legal persons9 hold 4060% of the 10 futures
varieties open interest,10 offering a reasonable market structure in
Figure 16.6 The historical ranking of Chinese futures exchanges among all
global exchanges
0
ZCE DCE SFE CFFEX
5
10
15
20
25
30
35
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
420
Table 16.4 Chinas international consumption status of its futures underlying

commodities
China positioned Futures underlying commodities Number

in the world
The biggest consumer Copper, aluminium, rebar, wire, rubber, PTA,

PVC, cotton, soybean N0.1, soybean meal,
soybean oil, strong gluten wheat, hard wheat,
rapeseed oil, palm oil, early long-grain
non-glutinous rice, lead, coke, methanol 20
The second biggest Gold, silver, LLDPE, sugar, corn 5

consumer
The third biggest Fuel oil, soybean No. 2 2

consumer
The second biggest Stock index futures 1

stock market
open interest.11 The percentages for aluminium, soybean No. 1, palm

oil, hard wheat and rapeseed oil are high above 60%, which is over-
high. The percentages for gold, rebar, rubber, LLDPE, soybean No. 2,
coke, sugar, early long-grain non-glutinous rice and stock index
future are below 40%, which is relatively low.
Individuals are the main trading participants, usually more than
half of total trading volumes. The volume percentage12 of individ-
uals for soybean No. 2 and rubber is over 80%; for gold, rebar,
LLDPE, coke, sugar, early long-grain non-glutinous rice, stock index,
zinc, PVC, strong gluten wheat, cotton and PTA is 5080%; for
aluminium, lead, fuel oil, copper, soybean N0.1, soybean meal,
soybean oil, corn, palm oil, hard wheat and rapeseed oil is below
50%.
Price volatility between futures and spot

The assessment shows that the price volatility between futures and
spot is almost at the same level, except for the price of strong gluten
wheat, hard white wheat and early long-grain non-glutinous rice
contracts, whose volatilities are slightly higher than the actual price
volatility. Research data reveal that price correlation between futures
and spot on the majority of products in Chinas futures markets are
stable and functional.
421
Declared hedging positions

Research shows that commercial enterprises mainly took short posi-
tions to hedge, based on the application for hedging quota, the use of
that hedging quota and the hedging of open interest. Based on the
actual situation in China, upstream enterprises often have a large
production capacity and therefore exhibit great enthusiasm for
futures market participation, while the downstream enterprises are
mainly small businesses, usually lacking hedging abilities and will-
ingness. This results in a large amount of short hedging. In the
futures market, the manufacturers naturally hold short positions,
which creates a downward pressure on futures price. As the counter-
party, speculators generally tend to hold long positions, hoping to
profit when the futures prices increase. Manufacturers pay a
premium to manage the risks coming from price fluctuations.
Price changes and price hit limits

Generally, the futures market prices ran smoothly in 2011, with the
hitting of the upper or lower price limits three times in a row
decreasing year-on-year.13 However, the futures prices of copper,
rubber, LLDPE, PVC, coke and palm oil experienced rather frequent
fluctuations, with each hitting the price limit more than 25 times.
LLDPE hit the price limit three times in a row on 16 occasions.
Short-term trading activity in the market

The short-term trading of futures was active, with intra-day trading
generally over 50% of total trading volume. In 2011, the three vari-
eties that generated the largest proportion of day trading volume
were rubber, cotton and stock index futures. The intra-day trading
activity for these contracts accounted for over 80% of the total. The
three varieties with the lowest proportion of intra-day trading were
soybean No. 2, corn and hard wheat, all of which had an intra-day
trading proportion below 50%.
ANALYSIS OF TRADING CHARACTERISTICS IN 2011

The trading volume of futures and options rose worldwide but
dropped in China
Global futures and options market trading volume increased
Statistics show14 that the global futures trading volume increased by
7.4% from 12.049 billion transactions in 2010 to 12.945 billion transac-
422
tions in 2011, and that global options trading volume increased by

15.9% from 10.375 billion transactions in 2010 to 12.027 billion trans-
actions in 2011. In 2011, the global futures and options trading
volume was 24.972 billion contracts, an increase of 11.4% from 22.425
billion transactions in 2010.
Futures trading volume in China decreased

According to the overall analysis from CFA data, in 2011 the trading
volume of the Chinas commodity and financial futures was 1.054
billion transactions, a decrease of 32.72% from 1.567 billion transac-
tions in 2010. Chinas proportion of commodity and financial futures
trading volume in the world decreased from 6.82% in 2010 to 4.22%
in 2011. Chinas commodity futures exchanges accounted for 53.65%
and 38.03% of the worlds futures trading volume for 2010 and 2011,
respectively.
It is worth noting that, despite an 11.36% growth in the global
futures and options markets trading volume in 2011, the global
commodity futures and options trading volume fell by 6%.
Agricultural commodity futures trading volume dropped by 26%
and non-precious metal futures fell by 33%. Although the national
futures trading volume declined by more than 30%, Chinas financial
futures trading volume grew in 2011. CFFEXs trading volume was
50.41 million transactions in 2011, an increase of nearly 10 % over
2010.
Chinas commodity markets still worlds largest

In FIAs Annual Volume Survey, ZCE ranked 11th in 2011 (12th
for 2010), SHFE ranked 14th in 2011 (11th for 2010) and DCE ranked
15th (13th for 2010). In 2011, the trading volume of ZCE decreased
only by 18%, while DCEs decreased by 28% and SHFE by 50%.
Among the 32 countries for which FIA collected data, there are
commodity futures markets in 25 of those countries/regions. There
are 45 commodity derivatives exchanges, with total trading volume
reaching 2.64 billion transactions. Despite a decline of 34.01%
between 2010 and 2011, the trading volume of Chinas commodity
futures in 2011 amounted to 1.003 billion contracts, accounting for
38.03% of the world volume and ranking first place globally. Figure
16.7 shows the proportion of total commodity future trading volume
by country.
423
Figure 16.7 Proportion of commodity futures (2011)
US 28%
China 38%
UK 15%
India 15%
Japan, Russia,
others 1%
respectively
Source: Based
d on FIA data
Of the global top 20 agricultural contracts by volume in 2011, nine

were in China. ZCE cotton futures ranked first and sugar ranked
second. The other seven contracts were from DCE (soybean oil
ranked fifth, soybean meal ranked sixth, corn ranked ninth, soybean
N0.1 ranked 10th and palm oil ranked 14th), SHFE (rubber ranked
third) and ZCE (strong gluten wheat ranked 20th). Moreover,
SHFEs rebar, zinc and copper futures ranked first, fifth and seventh,
respectively, among global top listing on metal derivatives.
Why the trading volume of futures and options rose worldwide

but dropped in China
The increase of the trading volume of futures and options globally
Below are some of the reasons behind the increase in trading
volumes.
Most of global futures exchanges experienced an increasing

trading volume: among the 77 exchanges around the world, the
trading volume of 59 exchanges grew (accounting for 77%), 14
exchanges grew by 50% (accounting for 18%) and 11 exchanges
grew by 100% (accounting for 14%).
Several exchanges experienced a rapid or steady increase in
trading volume: in 2011, the Singapore Mercantile Exchange, C2
Options Exchange in the US, BATS Exchange and Australian
Stock Exchange enjoyed the highest growth rate, India Joint
424
Stock Exchange achieved a growth rate of 181%, Thailand

Futures Exchange saw a growth rate of 136% and
IntercontinentalExchange (ICE) futures trading volume hit a
record high.
Precious metals futures contracts enjoyed a significant increase
ted to 0.179 billion

0. Worldwide, three
d futures contracts.
million globally, an
xchanges launched
15
d.
The decline in Chinas futures trading volume

In 2011, although Chinas futures market made significant achieve-
ment in improving the market structure, market function and futures
product innovation, the trading volume of the market suffered a
marked decline after five years rapid growth.16 There were three
reasons for this. First, the turmoil of the global economy, the struc-
tural transformation the domestic economy was facing and the
slowdown of the economic growth rate adversely impacted the
futures market. Second, factors such as the global loose-monetary
environment and the speculation of some commodities boosted the
price and volume of the futures market. Third, since the fourth
quarter of 2010 many measures have been taken to curb excessive
speculation, such as suspending the trading fee preferential system,
restricting the size of opening positions and raising the margin
requirement.
FUTURES COMPANIES AND TRADING VOLUMES

Futures companies
At the end of 2011, the margin held by the futures industry was
nearly US$25.59 billion. The various reserves that can be used to deal
with market risks was US$1.74 billion, out of which the risk reserve
of futures brokerage companies was US$304.98 million, the risk
reserve of future exchanges was US$1.09 billion and investor protec-
tion funds was US$345.10 million, which enhanced the industrys
ability to control risk. However, compared to other financial
425
institutions, futures brokerage companies overall strength was still

quite low. At end-2011, there were 161 futures brokerage companies
in China, and their total net capital was US$5.28 billion, only
US$32.10 million for each company on average. The net capital of the
top 10 futures brokerage companies only accounted for 28% of the
industry total, 32% for clients margin and 28% for fee income.
Futures brokerage companies income source was mainly brokerage
fee income and interest income from customers margin deposits. In
2011, the revenue of 161 futures brokerage companies was US$2.46
billion, out of which brokerage fees were US$1.60 billion and interest
income from customers margin was US$513.64 million. The annual
net profit was US$369.18 million, and some futures companies even
suffered a loss (data from CSRC).17
2012 trading volumes

In 2012, the total volume of futures market in China was 1.45 billion
contracts, with a notional value of approximately US$27.47 trillion,
increasing by 37.60% and 24.44% from 2011, respectively.
The trading volume of Shanghai Futures Exchange was 0.365
billion contracts, with a notional value of approximately US$7.12 tril-
lion, increasing by 18.52% and 2.63% from 2011, respectively, and
with a market share of 25.19% and 26.06%, respectively.
The trading volume of Zhengzhou Commodity Exchange was
0.347 billion contracts, with a notional value of approximately
US$2.79 trillion, decreasing by 14.61% and 48.04% from 2011, respec-
tively, and with a market share of 23.93% and 10.15%, respectively.
The trading volume of Dalian Commodity Exchange was 0.633
The trading volume of China Financial Futures Exchange was 0.105
THE FUTURE
The development of Chinas futures market
The year 2012 witnessed the 22nd anniversary of the establishment
and development of Chinas futures market. Since the early 1990s,
426
Chinas futures market has grown to become one of largest

commodity futures exchanges in the world on the basis of trading
volume. In 2009, the total trading volume of Chinas commodity
accounted for 43% of global volume, the biggest commodity futures
market globally.
There are three reasons for the prosperity of Chinas futures
market. First, Chinas futures market enjoys great support from
governmental policy. Since 1988, the establishment of a sound and
steady futures market has been included and referred to in the
governments working report six times. Since 2004, requirements
have been raised in the No. 1 central document18 on the futures
market that futures derivatives for commodities should be trans-
formed to a risk management instrument used by the real economy.
Second, the regulatory agency has strengthened the supervision of
the futures market to control risk. The regulatory agency always
gives top priority to strengthening the infrastructure of the futures
market. In addition, a good environment has been created for the
development of the futures market through continuously reinforcing
the basic regulations and institutions, steadily promoting new prod-
ucts, enhancing frontline supervision and strictly preventing market
risk.
Third, Chinas futures market is generally managed well, not only
attracting a number of commercial enterprises to hedge, but also
Figure 16.8 Chinas futures transactions and GDP

1600 60.00
Contracts in millions
GDP in trillions
1400
50.00
Contracts in millions
1200
GDP in trillions
40.00
1000
800 30.00
600
20.00
400
10.00
200
0 0.00
1993 1995 1997 1999 2001 2003 2005 2007 2009 2011
427
increasing the integration of the futures market, cash market and

industrial entities. Also, the key reason is that the futures market
keeps up with Chinas rapid economic development and soaring
commodity cash market.
Huge development space of Chinas futures market

Although the trading volume of Chinas futures market tops any
other, some development-restricting factors cannot be ignored. For
example, the futures market is faced with an environment of exces-
sive regulatory restrictions, community monopoly,19 OTC disguised
futures,20 over-speculation and the problem of a lack of creativity in
development, competitiveness in exchange and low level of diversity
of futures varieties. While the domestic futures market is active, the
international pricing power is still out of Chinas control.
Additionally, the monotonous operating system, improper regula-
tory system and single futures agent business21 have held back the
development of the futures market. It should be admitted that
Chinas futures market has a long way to go. For instance, cash settle-
ment of commodities has not been implemented and cross-product
arbitrage trading, transactions after closing (post-market session
trading), margin netting, SPAN, exchanges of futures for swaps and
options trading have not been instigated.
It is not yet possible to determine the steps necessary to achieve
these innovative projects or how long it will take and how difficult it
will be. To improve the international status of Chinas financial
sector, a strong futures market is needed for support. To improve
Chinas international competitiveness, the price discovery function
of futures markets should be given full play; to cope with the impact
from the international financial crisis, the hedging function of the
futures market should be further promoted; to improve the national
economy growth, service level of the futures markets should be
improved. Therefore, China enjoys a huge development opportunity
in the future.
Possible changes to Chinas futures market

In spite of the high trading volume, the operating quality of Chinas
futures market is not completely satisfactory. Since the financial
crisis of 200708, the global financial environment has changed
constantly, and the futures market has been confronted by great
428
uncertainty, offering both risks and opportunities. However, since

the early 1990s, Chinas futures market has gone through stages of
blind development, rectification and regulatory change, and come to
a critical period of development, namely a development process
from quantitative change to qualitative change.
Undoubtedly, through this process some essential changes will
occur in Chinas futures market. The first is that the futures market
will operate within the rule of the law, although there is still no
national futures law in China. Second, with the increasing number of
futures varieties, the coverage of futures products will be wider, and
futures markets should tightly track the actual underlying markets.
In addition, a speculative component will be more suitable for the
operation of the market. Third, there will be an increasing number of
innovations in the market-operating mechanism. The investor struc-
ture will be optimised and the proportion of institutional investors
will be raised. Fourth, with the improvement of international
competitiveness, the pricing power of the futures market will be
strengthened. Finally, new ways should be constantly opened up to
prevent market risks.
THREE MAJOR DEVELOPING TRENDS OF CHINAS FUTURES

MARKET
The trend to loosen product-listing control
In order to curb the excessive amount of different types of futures
that were listed by the exchanges in the early days of the futures
market in the 1990s, an approval mechanism for the listed futures
varieties has been implemented. This prevents premature futures
varieties from listing, prevents the futures market from deviating
from the real economy, ensures the consistence of the futures and
cash market, controls market risk and standardises the market order.
However, under the approval mechanism, despite strong efforts for
promoting new products by the regulators and exchanges, a new
product often has a slow introduction process and may miss the best
time to market owing due to cumbersome procedures.22 This indi-
cates that the inefficient administrative approval mechanism fails to
keep pace with the needs of the market and may hinder the pricing
function of the futures market. Thus, futures markets do not always
provide the risk management tools needed by all enterprises. In
addition, in the majority of futures exchanges around the world,
429
nearly all varieties of agricultural futures have corresponding

futures options trading to allow flexible hedging of the agricultural
market risks. However, options trading has not appeared uniformly
in Chinas futures markets, preventing domestic investors from
taking a more flexible approach to avoid market risks and also
restricting enterprises from participating in hedging. Therefore, in
the future, Chinas futures market should relax product control and
gradually allow more listing of new products.
Corporatisation of the exchanges

Chinas futures exchanges are structured uniquely, and their owner-
ship is ambiguous and confusing, even to professionals in China.23
Although, in economic terms, according to the Regulations on
Administration of Futures Trading, 24 exchanges are seen as corpo-
rate legal persons or other economic organisations, in reality, in
terms of management, the institutional arrangements stipulated by
the Futures Exchange Management Regulations25 implies very
strong administrative factors. Therefore, the exchanges that act as
administrative organisations should be institutional legal persons.
On the other hand, organisationally, judging by the source of their
capital and their stated aims,26 exchanges that have features of social
community should be seen as social group legal persons (legal
persons are classified under several different categories in China).
Therefore, exchange systems should come under a special organisa-
tional model.27 In the authors opinion, since the futures market is
market-oriented, complex and has high risk, it should be dynami-
cally regulated for specialisation and legalisation, and even the
regulatory system itself should be market-oriented. The trend of the
corporatisation28 of exchanges around the world proves that corpo-
ratisation does not contradict the unified regulation of the futures
market by the government. Therefore, with the further opening-up
of Chinas futures market as a part of its regulatory system,
exchanges should be equipped with a more advanced independent
market development force and much stronger competitiveness. As a
result, corporatised exchanges will become a development trend of
Chinas futures market.
430
The further opening up of the futures market

Since 2008 (and especially since 2010), the internationalisation of the
Chinas futures market has been a hot topic in the domestic futures
industry. A consensus has been reached, that it is the basic strategy of
Chinas economic development to insist in opening up to allow
active participation in the international market. Since the futures
market directly serves the national economy, its internationalisation
is both an internal need for the development of Chinas economy and
an irresistible development trend. There are two reasons to
explaining this. On one hand, our enterprises competing for the
pricing power of commodities requires the international futures
companies to provide intermediary services, and the prices of inter-
national commodities are based on the prices of international
commodity futures prices. Only by actively participating in the same
markets with the leading international futures companies and grad-
ually expanding its influence, can Chinas enterprises overcome the
difficulty of gaining pricing power. On the other hand, international-
isation can enhance the competitiveness of the futures companies. In
the process of internationalisation, in addition to domestic brokerage
business, futures companies can expand their businesses to foreign
business, as well as business that combines domestic and foreign
operations, which can improve operations and competitiveness. As
predicted, a series of opening-up measures, including product cross-
listing, stock holding of foreign investors and international product
listings will become one of the main developing trends of Chinas
futures market.
SUGGESTIONS
Stable trading allows for progress while ensuring stability
Influenced by factors such as state macro-control and complicated
market conditions, there has been a large fluctuation in the trading
volume of Chinas futures market, especially in 2008. Chinas futures
trading volume increased to 87.24% as a result of the international
financial crisis, before the rate decreased again by 20.72% in 2009. In
2010, it rocketed up again to 84.74%, but then sharply declined to
30.69% negative growth in 2011. On the contrary, since the early
2000s the proportion of US futures and options trading volume in the
world increased by more than 30%. Since the financial crisis, the
trading volume of the US market has gone up steadily. In 2011, its
431
futures and options trading volume was 8.119 billion contracts,

14.01% more than the 7.121 billion contracts of 2012, occupying
32.51% of the total world shares, a small increase on the 31.94% of
2010. Thus, compared with the futures market of the US and other
developed countries, the stability of Chinas futures trading volume
should further improve. Therefore, in a situation where futures
trading volume might decline, the relation between trading volume
and upgrading the quality of the markets should be handled
correctly and the scale of the futures market should be kept relatively
stable, both of which are not inconsistent with the principle of
making progress while ensuring stability of the Central Economic
Working Conference.29 That is to say, it would promote the transfor-
mation of Chinas futures market from quantity expansion to quality
expansion.
The insurance function and development of options

International experience shows that, since the early 2000s, the global
options market has developed quickly. Statistics show that the
trading volume of options outpaced futures, apart from in 2010 and
2011. However, as the data of different classes of futures shows, the
trading volume of stock index option was several times that of stock
index futures, but commodity options trading volume was only one-
tenth of the futures volume.30
Figure 16.9 Global options volume as percentage of futures

12%
10%
8%
6%
4%
2%
0%
Global commodities Global energy Global agricultural Global metals
Source: http://www.futuresindustry.org
432
As shown in Figure 16.9, in 2011 global energy options transac-

tions accounted for 11% of its futures trading volume, global
agricultural options trading was 7.88% of the agricultural futures
and global non-precious metals and precious metals options trading
volume accounted for only 2.8% of the futures trading volume.
It can be seen that the original intention and purpose of listing
commodity options was mainly to serve as an insurance function for
futures and to enhance the quality of the futures market, not to
increase trading volume. For this reason, it is recommended that a
further study on the insurance function of options for futures and a
project on options training including simulated options trading is
needed.
CONCLUSIONS
For the optimal development of Chinas futures market, the
following should be carried out:
further research and development of the crude oil futures;

rules of Treasury bond
433
overall operations, despite the growth rate having slowed. Under

these conditions, the development of the real economy needs a rapid
development of Chinas futures market through greater specialisa-
tion, diversification and internationalisation. It offers the futures
markets a historic opportunity, but also some serious challenges.
Chinas futures market needs not only a stable and functioning
market operation, but also much more innovation and opening up to
the world. This is the common aspiration for people in the industry
and is also the objective requirement for economic development in
China. The Chinese market has a great deal of room to develop, with
more new products to be listed, more market innovation and to be
more involved in the worlds derivatives business.
Finally, the question remains whether it is possible that these
markets will open to foreign traders. There is no schedule for foreign
investors to start trading in Chinas commodity markets, but that time
is getting closer every day. Regulators are taking much more interest
in allowing foreign investors to trade in domestic markets, as
researched and provided by futures exchanges. Regulators are even
guiding exchanges to research some of the programmes aimed at
attracting foreign investors. In 2012, Chinas regulators implemented
a series of measures to speed up the opening of the capital markets to
the outside world, including the qualified foreign institutional
investor (QFII) and renminbi qualified foreign institutional investor
(RQFII) systems. In 2012, 72 QFIIs were approved. At the time of
writing in 2013, a total of 213 qualified QFIIs had gained an approved
investment quota amounting to US$39.985 billion. At the same time,
in order to further expand opening to the capital market, in September
2012 the Shanghai and Shenzhen stock exchanges, as well as the China
financial futures exchange, held several QFII promotional activities in
the US, Canada, Europe, Middle East, Japan and South Korea to intro-
duce foreign investors to Chinas capital market. Furthermore, the
Shanghai futures exchanges crude oil futures are expected to be intro-
duced in 2013, with the aim of attracting more international investors
to participate in the financial and capital market in China.
1 Zhengzhou Grain Wholesale Market is the predecessor of Zhengzhou Commodity

Exchange. It is located on the south bank of the Yellow River, and is the first experimental
futures market in mainland China. Henan is a large agricultural province, whose output of
grain, cotton and oil rank among the top in China. It is an important base of quality agricul-
tural products.
434
2 Natural rubber R708, China Commodity Futures Exchange, Inc of Hainan (CCFE) in 1997;
Coffee F605, F607, F609, F703, China Commodity Futures Exchange, Inc of Hainan (CCFE)
in 1996-1997; Jun Plywood 9607, Shanghai Futures Exchange (SHFE) in 1996; Red bean,
Suzhou Commodity Exchange in Jan-Mar 1996; Soybean meal 9601, 9607, 9708 Guangdong
United Futures Exchange (GUFE) in Oct-Nov 1995; Sticky rice 9511, Guangdong United
Futures Exchange (GUFE) in Oct 1995; Red bean 507, Tianjin Commodity Exchange in May-
Jun 1995; 3-year T-bond 314, 327, Shanghai Securities Exchange (SHSE) in 1995; Palm Oil
M506, China Commodity Futures Exchange, Inc of Hainan (CCFE) Mar 1995; Corn C511,
Dalian Commodity Exchange (DCE)Futures in 1995; Steel wire, Suzhou Commodity
Exchange in 1994-1995; Japonica rice, Shanghai Food and Oil Exchange in Jul-Oct 1994.
3 Including a serial of documents, such as Provisional Regulations on Futures Trading,
Futures Exchange Management Regulations, Futures Brokerage Corporate Management
Approach, Futures Brokerage Companys Senior Management Personnel Qualifications
Management Approach and Futures Industry Practitioners Qualified Management
Methods.
4 In the early 1990s, the Zhengzhou Commodity Exchange listed mung bean futures, the first
batch of domestic listed contracts. In 1998 and 1999, for two consecutive years it traded at a
peak of more than half of the national futures trading volume. In late 1999, after the margin
increased from 5% to 1520%, the mung bean futures had no trades. In May 2009, after the
CSRC approved the Request for Suspension of Mung Bean Futures Trading, which was
delivered by ZCE in 2008, the mung bean contract was delisted.
5 On June 2, 1999, the State Council issued the Interim Regulations on Administration of
Futures Trading. After it was revised, the new Regulations on Administration of Futures
Trading took effect on April 15, 2007. The newest Regulations on Administration of
Futures Trading were approval by the State Council on September 12, 2012, taking effect on
December 1, 2012.
6 China Futures Association consists of group members, including members of the futures
brokerage firms, special members of futures exchanges and individual members in the
futures industry. It accepts business guidance and management from the China Securities
Regulatory Commission and the Organization of National Social Group Registration
Administration.
7 http://futures.hexun.com/20120515/141434876.html.
8 According to the internal research materials of regulators (and the below is the same).
9 Refers to an individual or group that is allowed by law to take legal action as plaintiff or
defendant. It may include natural as well as fictitious persons (such as corporations).
10 Includes lead, fuel, copper, zinc, PVC, soybean meal, corn, strong gluten wheat, cotton,
PTA.
11 Refers to the ratio of the position of legal person to the total position in the related contract
market.
12 Refers to the ratio of individuals trading volume to total trading volume in the related
contract.
13 For example, if the accumulative price increase (decrease) based on settlement price in three
successive trade days (D1, D2, D3) of a futures contract reaches three times of the stipulated
increase (decrease) price limit in the contract, the exchange has the right to increase the
margin rate with the scale no more than three times of the trading margin standard in the
contract applicable at that time. Under circumstances of significant increased market risk
caused by the special situation of some futures contract, the exchange may take the
following measures according to the market risk of some futures contract, such as putting
limitations on margin movements, putting limitations on opening position and closing posi-
tion, adjusting the trading margin standard of this futures contract and adjusting the range
of price limits of the contract.
14 Analysis and calculation based on data from the Futures Industry Association.
435
15 http://www.tygedu.com/newsviews.php?newsid=4234.
16 Annual growth rate from 2006 to 2010 is, respectively, 32.22%, 62.06%, 87.24%, 20.27% and
84.74%.
17 http://www.pbc.gov.cn/image_public/UserFiles/goutongjiaoliu/upload/File/%E4%
B8%AD%E5%9B%BD%E9%87%91%E8%9E%8D%E7%A8%B3%E5%AE%9A%E6%8A%A5
%E5%91%8A%EF%BC%882012%EF%BC%89.pdf.
18 No. 1 central file refers to the first document issued by the Central Committee of
Communist Party of China every year. It has become the proper noun for showing that
the Central Committee of Communist Party of China attaches great importance to rural
problems.
19 The monopoly behaviour of the futures industry in China is mainly reflected in the admin-
istrative monopoly, futures exchanges monopoly and futures brokerage industry
monopoly. For example, the regulatory department dominates the approval of varieties of
futures and products, listing locations as well as restricting every product to be listed on one
exchange without saying that listed procedures are cumbersome. In addition, in relation to
a futures company, an exchange has a much stronger voice and negotiation ability, etc.
20 In China, outside of futures exchanges, there are more than 200 electronic commodity
trading markets that trade almost the same contracts and use trading mechanisms and rules
similar to futures exchanges. This is called disguised futures or quasi futures. The
markets lack strict regulation and market rules, and their operation and implementation are
in chaos. To some extent, they inhibit the healthy development of Chinas futures markets.
21 Since the early days, risk events have occurred with futures companies misappropriating
customer funds and allowing customers to carry out overdraft trading. However, futures
brokers firms are not allowed to trade their own accounts and can only trade for customer
accounts, promulgated and provisioned by the regulations on futures trading of September
1999. The main business of futures companies in China is to trade on behalf of their
customers, so the firms income mainly comes from commission.
22 The listing mechanism for new futures contracts in China adopts a non-market-approval
system. In order to list a new variety of futures contract, it must first undergo a repeated
research and review process by the futures exchange and then report to the CSRC accord-
ingly. After examination and verification, CSRC then submits a report to the State Council,
which needs to consult the relevant state ministries and commissions, related industry
administration departments and even relevant provinces and cities. After getting all feed-
back, the State Council might make written instructions. A new variety of futures contract
will finally be launched.
23 Chinas three commodity futures exchanges are institutional units implementing enterprise
management, while China Financial Futures Exchange is a corporate system, an enterprise
is a legal person. In fact, they are all to different extents the subsidiary of the regulatory
department.
24 Article 7: non-profit futures exchanges shall carry out self-regulation according to their
constitutions, and assume the civil liabilities with all their properties. A futures exchange
bears civil liabilities to the extent of all of its property. The persons in charge of futures
exchanges shall be appointed and dismissed by the futures supervision and administration
authority of the State Council. Article 18: The revenues obtained by a futures exchange shall
be managed and used in accordance with the relevant state regulations and may not be
distributed to members or diverted for other uses.
25 A futures exchange shall establish a board of directors. The chairman and the vice-chairman
shall be nominated by the CSRC and elected by the board.
26 Article 4: Registered capital of the membership futures exchange is divided into equal
shares, which is subscribed to by its members.
27 This is not only difficult to explain to most Westerners, but also to some Chinese.
28 The meaning is similar to demutualisation ie, the move from a member-owned organi-
sation to a publicly listed organisation.
436
29 The Central Economic Working Conference of the Central Committee of the Communist
Party and the State Council is the highest economic conference in the nation. Its mission is to
carry out economic achievements, deal with international and domestic economic changes,
formulate macroeconomic development planning and deploy the following years economic
work.
30 Data source from FIA statistics and calculated by author (the below is the same).
REFERENCES
Falloon, William D., 1998, Market Maker: A Sesquicentennial Look at the Chicago Board of
Trade (Chicago, Ill: CBOT).
Gorham, Michael and Nidhi Singh, 2009, Electronic Exchanges: The Global Transformation
from Pits to Bits (Burlington, MA: Elsevier).
Hieronymus, Thomas A., 1996, A Revisionist Chronology of Papers.
Hieronymus, Thomas A., 1997, Economics of Futures Trading: For Commercial and
Personal Profit, Commodity Research Bureau.
Melamed, Leo, 1993, Leo Melamed on the Markets: Twenty Years of Financial History as Seen by
the Man Who Reolutionized the Markets (New York, NY: Wiley).
Ronalds, Nick and Wang Xueqin, 2006, The Dawning of Financial Futures in China,
Futures Industry, November/December.
Ronalds, Nick and Wang Xueqin, 2007, China: Futures Take Dragon Steps, Futures
Industry, September/October.
Ronalds, Nick and Wang Xueqin, 2007, Chinese Commodity Markets: History,
Development and Prospects, in Hilary Till and Joseph Eagleeye (Eds), Intelligent
Commodity Investing: New Strategies and Practical Insights for Informed Decision Making
(London: Risk Books).
Xueqin, Wang, 2008, Research on Options on Commodity Futures (China Financial &
Economic Publishing House).
Xueqin, Wang and Michael Gorham, 2002, The Short, Dramatic History of Futures
Markets in China, Journal of Global Financial Markets, Spring, pp 2028.
Xueqin, Wang and Nick Ronalds, 2005, The Fall and Rise of Chinese Futures 19902005,
Futures Industry, May/June.
437
17 Index CIT_Commodity Investing and Trading 30/09/2013 14:41 Page 439
Index
(page numbers in italic type refer to figures and tables)
A and options volatility 292

agriculture: markets, trading in 26179
and farmland as investment, see and correlation benefit 2658,
farmland 267, 269, 271, 272
and impact of non-fundamental fundamental data points
information on commodity 2738
markets 79, 8, 9 and investment flows,
market participants 24960 seasonality and weather
commercial traders 24955, 26873
251, 254 market environments and
commodity index and swap volatility 2612
traders 260 and strategy selection 2625,
non-commercial traders 266
2556 technical inputs 2789
proprietary traders: individual and mollisols 230
and trading groups trading in 24994, 251, 252, 254,
2568 255, 257, 263, 264, 266, 267,
systematic and technical 269, 271, 272, 274, 275, 276,
traders 2589 277, 280, 281, 2845, 286,
market strategies in 27992 288, 289, 290, 291, 293
calendar spreads 2837, 2845, and weather 66, 723
288 see also farmland
crush spreads 28992, 291 American Petroleum Institute
directional 2823 103
geographical spread arbitrage Anderson, Dwight 385
2879, 289 APX-ENDEX 119
options 293 Australian Stock Exchange 424
439
B 413, 415, 416, 418, 419, 420,

Barnett Shale 368 421, 422, 424, 427, 432
base metals: clean-up and rectification, first
exchanges that list 1445, 144 phase of 41012
price movement in 6 clean-up and rectification,
BATS Exchange 424 second phase of 41214
BHP Billiton 373 companies 4256
BP 91 and corporatisation of
Brazil: exchanges 430
and coffee prices 74 development of 4267
growing economic prominence and exchanges,
of 86 corporatisation of 430
as major meat exporter 177 and financial futures,
see also BRIC economies successful launch of 418
Brent 45, 5, 14, 935, 94, 979 first ten years (19902000) 410
Dated 94, 978 and foreign capital 417
BRIC economies 105 further opening up of 431
see also oil and petroleum and futures and spot, and
products: historical price price volatility between 421
perspective on and hedging positions,
Brookings Institution 245 declaring 422
and Hong Kong branches of
C future commission
C2 Options Exchange 424 merchants (FCMs) 417
capacity allocation and congestion huge development space of
management (CACM) 119 428
CBOT, see Chicago Board of Trade and insurance function and
CFFEX, see China Financial Futures development of options
Exchange 4323
Chvez , Hugo 106 major developing trends
Chevron 83 42931
Chicago Board of Trade (CBOT) market operation, micro-
166, 184 characteristics of 420
Chicago Mercantile Exchange and open interest 4201
(CME) 27, 89, 101, 21415, and opening of futures
250, 260 companies 417
China: possible changes to 4289
and farmland 2424, 243 and price changes and price
futures market of 40937, 411, hit limits 422
440
INDEX
and product-listing control, regulation 21520

loosening 42930 consumption 21720
and promotion of new futures regulation 21516
contracts 41617 supply and trade 216
regulatory regime, terminology and conversion
determining 41415, 415 statistics 2235
second ten years (200010) and transportation 21113
41518 Comex Copper Exchange 144
and short-term trading activity commodities index investing
422 33941
and stable trading and Commodity Futures Trading
stability 4312 Commission (CFTC) 249
trading characteristics 2011, commodity markets:
analysis of 4225 coal 20725, 209, 216, 217, 218,
trading volumes (200010) 219, 223
41820 characteristics of 20813, 209,
trading volumes (2012) 426 210, 21011
increased number of refineries in conversation statistics and
99 terminology 2235
as key driver in metals 137 financial markets for 21415,
and LNG imports 60; see also 215
BRIC economies and fuel-to-power spreads
MMT growth of 56 2201
net oil consumption of 107 market structure for 21314
China Financial Futures Exchange overview 2078
(CFFEX) 416, 418, 426 and supply, demand and
coal 20725, 209, 216, 217, 218, 219, regulation 21516, 21520,
223 216, 21720
characteristics of 20813, 209 terminology and conversion
metallurgical 21011 statistics 2235
thermal 210 and transportation 21113
conversation statistics and and energy and commodity
terminology 2235 physical and financial
financial markets for 21415 portfolios, enterprise risk
CME product slate 215 management for, see main
and fuel-to-power spreads 2201 entry
market structure for 21314 excess capacity enjoyed by 3
overview 2078 farmland 22947; see also
and supply, demand and farmland
441
grains and oilseeds 165206, 169, nickel 1534

170, 172, 175, 180, 182, 191, palladium 164
194, 195, 202 physical factors driving
and agribusiness investors, market in (listed) 148
listed 206 platinum 163
feed, food and vegetable precious 1457
proteins 166201; see also prices 13944, 139, 140, 141,
under grains and oilseeds 142, 143
and genetic modification 166, scrap 138
240 silver 1612
impact of non-fundamental supply 1368
information on 324, 6, 7, 8, zinc 1556
9, 10, 11, 12, 13, 14, 15, 16, North American natural gas
17, 18, 19 2564, 26, 28, 29, 31, 32, 34,
and agriculture 79, 8, 9; see 35, 37, 38, 39, 40, 45, 48, 50,
also farmland 51, 55; see also natural gas
and base metals 6 and coming decade, key issues
and Dow Jones- UBS (DJUBS) for 53
1013, 202 and demand-side dynamics
and energy 45, 5 2836
increased influence of 4 and futures, price dynamics in
and precious metals 6 4953
and sentiment 920, 10, 11, 12, geography of production and
13, 14, 15, 16, 17, 18, 19 demand 47
and SG sentiment indicator measures and conversions,
911 common units of 26
in metals 13364, 135, 136, 137, overview 256
138, 139, 140, 141, 142, 143, and regasification 58
144, 14950, 1512, 1534, storage 436, 45, 47, 48
1556, 157, 158, 15960, oil and petroleum, history and
1612, 163, 164; see also metals fundamentals 75110; see
aluminium 14950 also oil and petroleum
base 1445 products
bulk 1456, 158 putting momentum into 32535
copper 1512 and better strategy, building
demand 1356 3303
gold 15960 and commodity beta 326
inventory 1345, 135 and downside protection
iron ore 157 3345
442
INDEX
and long and short 327, 335 principal component analysis

and roll yield and excess conducted on 4
return 330, 330 strong gains in xvii
and sources of excess return commodity risk premiums
3279 295305, 298, 300, 301, 303
and sentiment 920, 10, 11, 12, active strategies 2967
13, 14, 15, 16, 17, 18, 19 benchmarks and overview 2967
in risk-off versus risk-on convergent and divergent
environments 1920, 19, 21, strategies 297304, 298, 303
22, 23 active example 3014
structural shift in 1218 Conservation Reserve Program
and SG sentiment indicator 911 168, 241
and structural alpha strategies, credit valuation adjustment (CVA)
see main entry 391, 392, 393, 395, 396, 397,
and weather 6574 398, 401, 402, 403, 404, 405,
data basics 667 406
day in the life 6770 and active risk management
impacts: on agriculture 723 with CVA desks 4067, 406
impacts: on livestock 734 allocation approaches 4003
impacts: on softs 74 and credit-adjusted rate curves
impacts: on transport 73 391
tropical conditions 702 and credit limits, charges and
in wholesale power 11331; see CVA desks 4035
also wholesale power CVA methods 3925
markets and debt valuation adjustment
and coming decade, key issues 390
for 12830 for energy and commodity
electricity 11418 derivatives 389407
and hedging strategies and explained 38990
price formation 1212 from portfolio CVA to deal CVA
and historical price 398400
perspective 1228 crude grades and locations 7983
sources of information see also oil and petroleum
concerning 1301 products
trading of, European markets CVR Energy 91
11821
see also specific commodities D
commodity prices: Dalian Commodity Exchange 166,
in metals 13944 414, 416, 417, 418, 426
443
Dated Brent 94, 978 and debt valuation adjustment

day-ahead and intraday markets 390
11920 explained 38990
Department of Agriculture (US) from portfolio CVA to deal
174, 231 CVA 398400
DoddFrank Act 130, 389 energy and commodity physical
Dow Jones-UBS (DJUBS) 10, 1013, and financial portfolios:
10, 11, 12, 23 enterprise risk management for
simple overlay example 202 363, 375, 376, 378, 379, 386
Dubai Mercantile Exchange (DME) backtesting 3802
27, 99, 100, 146 infrastructure 3835
and Ospraie Management
E LLC, risk management
electricity 11418 lessons from 385
market design 11718 policies and governance 3714
unique characteristics of 11417 risk-adjusted performance
dispatch arrangements 11416 measurement 383
locational issues concerning and risk management systems
11617 and data 3845
and quality 117 stress tests 37780
and scientific laws 114 valuation models for physical
see also commodity markets; assets, contracts and
wholesale power markets financial derivatives 3823
energy and commodity and valuation and risk
derivatives: methodologies and metrics
credit valuation adjustment 37483
(CVA): at-risk metrics: cashflow at
and active risk management risk, VaR and earnings at
with CVA desks 4067, 406 risk 3757, 375, 376
allocation approaches 4003 energy indexes:
and credit limits, charges and tracking 33764, 350, 3523,
CVA desks 4035 3567, 362, 363
credit valuation adjustment benchmark energy index, spot
(CVA) for 389407, 391, 392, and energy data 3469
393, 395, 396, 397, 398, 401, differential evolution
402, 403, 404, 405, 406 algorithm 361
and credit-adjusted rate evolutionary solution
curves 391 techniques 3446
CVA methods 3925 genetic algorithm 3613
444
INDEX
industry classification Shanghai Futures (SHFE) 144,

benchmark 3634 146, 414, 416, 417, 418, 423,
innovative approach 3416 424, 426
and problem formulation Shanghai Securities 418
3424 Shenzhen Securities 418
and SEI performance 34959 Singapore Mercantile 424
and selective portfolios, Thailand Futures 425
statistical properties of Zhengzhou Commodity (ZCE)
3559, 3567 414, 416, 417, 418, 423, 424,
Energy Information 426
Administration (EIA) 78 Exxon 105
Energy Market Observatory 130
EPEX Spot 119 F
Essar Oils 99 farmland:
European Network of bio-fuels from 231
Transmission System cash returns from 232
Operators (ENTSO) 130 and debt levels 231
exchanges: and declining worldwide
Australian Stock 424 inventories 231
BATS 424 defined 229
C2 Options 424 and food demand 231
Chicago Mercantile (CME) 27, versus population growth 245,
89, 101, 21415, 250, 260 245
China Financial Futures (CFFEX) and global farming 2415, 242
416, 418, 426 China 2424, 243
Comex Copper 144 demand for crops and
Dalian Commodity 166, 414, 416, commodities 2445
417, 418, 426 and inflation hedge 231
Dubai Mercantile (DME) 27, 99, as investment 22947, 233, 235,
100, 146 236, 239, 242, 243, 245
India Joint Stock 4245 and mollisols 230
Intercontinental (ICE) 27, 94, 101, and production, limitations of
214, 215, 425 2401
London Metal (LME) 144, 145 and conservation programmes
Minneapolis Grain (MGE) 184 241
Multi Commodity (MCX) 146 and renewables, impact of 23640
New York Mercantile (Nymex) agricultural products 23840;
27, 68, 69, 70, 71, 146, 299, 346 see also agriculture
New York Stock 364 fuels 2368
445
and resource conservation 231 storage 198

scarcity of 231 trends and swing factors for
as sustainable asset 232 future 2045
and US infrastructure 231 Great Britain:
and value creation and and wholesale power markets,
investment 2326 historical price perspective
and farmland as lease 1246, 125
property 2323 Gulf War, first 104, 107
and riskreturn profile 2334, see also oil and petroleum
233 products: historical price
see also agriculture perspective on
forward curves and market value
of storage 30911, 310 H
forward markets: Henry Hub 27, 62, 346
physical versus financial 120; see high-fructose corn syrup (HFCS)
also wholesale power 179
markets Hubbert, M. King 103, 105, 108, 110
products 1201 Hurricane Ike 92
Hurricane Katrina 42
G Hurricane Rita 42
gas futures: Hurricane Sandy 76
price dynamics in 4953; see also
natural gas I
see also commodity markets ICE, see IntercontinentalExchange
Germany: India:
and wholesale power markets, Jamnagar complex in 99
historical price perspective and LNG imports 60; see also
1224, 123 BRIC economies
grains and oilseeds 165206, 169, meat consumption in 1778
170, 172, 175, 180, 182, 191, MMT growth of 56
194, 195, 202 India Joint Stock Exchange 4245
feed, food and vegetable IntercontinentalExchange (ICE) 27,
proteins 166201 94, 101, 214, 215, 425
feed 17083 International Financial Reporting
food 18395 Standards (IFRS) 121
vegetable proteins (oilseeds) IranIraq war 103, 105
196201 see also oil and petroleum
and genetic modification 166, 240 products: historical price
and rotation 2014 perspective on
446
INDEX
Iranian revolution 103 calendar spreads 2837, 2845,

see also oil and petroleum 288
products: historical price crush spreads 28992, 291
perspective on directional 2823
Iraq, Kuwait invaded by 104; see geographical spread arbitrage
also oil and petroleum 2879, 289
products: historical price options 293
perspective on and options volatility 292
Markets in Financial Instruments
J Directive (MiFID) 12930
Jamnagar 99 metals 13364, 135, 136, 137, 138,
Japan, MMT growth of 56 139, 140, 141, 142, 143, 144,
JBS 199200 14950, 1512, 1534, 1556,
157, 158, 15960, 1612, 163,
K 164
Kuwait, Iraq invades 104; see also aluminium 14950
oil and petroleum products: base:
historical price perspective exchanges that list 1445,
on 144
price movement in 6
L bulk 1456, 158
Lehman Brothers 3, 4 copper 1512
liquefied natural gas (LNG) 549, demand 1356
59, 60, 61 gold 15960
exports of 548 inventory 1345, 135
future flow considerations for 63 iron ore 157
global 5464 nickel 1534
imports of 59, 60 palladium 164
imports of, and import growth physical factors driving market
5861 in (listed) 148
supply chain 623 platinum 163
London Metal Exchange (LME) precious 1457
144, 145 exchanges that list 146
gold among most actively
M traded 146
market information, sources of gold market an outlier among
1301 6
market strategies in agriculture many trading centres for 146
27992 most actively traded 146
447
prices 13944, 139, 140, 141, 142, and futures, price dynamics in
143 4953
scrap 138 geography of production and
silver 1612 demand 47
supply 1368 measures and conversions,
zinc 1556 common units of 26
meteorology, see weather overview 256
Minneapolis Grain Exchange and regasification 58
(MGE) 184 storage 436, 45, 47, 48, 49, 52
Mobil 105 peak in (2011) xvii
mollisols 230 supply-side considerations of
Multi Commodity Exchange 3643
(MCX) 146 and ethane rejection 423
shale 3642, 38
N weather impacts on 42
National Hurricane Center 70 Natural Resource Conservation
National Weather Service, US 66, Service (NRCS) 230
67 New York Mercantile Exchange
natural gas: (Nymex) 27, 68, 69, 70, 71,
demand-side dynamics for 146, 299, 346
2836 New York Stock Exchange 364
exports 346, 46 New York Times 72, 74, 104, 207
industrial use 2830, 28, 29 non-fundamental information:
power generation 304, 31, 32, impact of, on commodity
34 markets 324, 6, 7, 8, 9, 10,
residential and commercial 11, 12, 13, 14, 15, 16, 17, 18,
demand 34, 35, 37 19; see also commodity
futures, price dynamics in, and markets
futures, price dynamics in and agriculture 79, 8, 9
4953 and base metals 6
liquefied (LNG) 25, 34, 36 and Dow Jones- UBS (DJUBS)
global 16, 55, 57 1013, 202
North American market in and energy 45, 5
2564, 26, 28, 29, 31, 32, 34, increased influence of 4
35, 37, 38, 39, 40, 45, 48, 50, and precious metals 6
51, 55 and sentiment 920
and coming decade, key issues and SG sentiment indicator
for 53 911, 10, 11, 12, 13, 14, 15,
and demand-side dynamics 28 16, 17, 18, 19
448
INDEX
North American market in natural and crude pricing and trading

gas 2564, 26, 28, 29, 31, 32, 8997
34, 35, 37, 38, 39, 40, 45, 48, and crude production trends
50, 51, 55 since 1960 104
and coming decade, key issues and crude transport and choke
for 53 points 879, 87, 88
and demand-side dynamics 28 historical price perspective on
and futures, price dynamics in 10110
4953 and OPEC 1016 passim
geography of production and major supply, locations of 837;
demand 47 see also oil and petroleum
measures and conversions, products: and crude grades
common units of 26 and locations
overview 256 reason for oil 75101
and regasification 58 and seasonality 789, 80
storage 436, 45, 47 storage 8990
Notice to Resolutely Stop the oilseeds and grains 165206, 169,
Blind Development of the 170, 172, 175, 180, 182, 191,
Futures Markets (China) 194, 195, 202
410 feed, food and vegetable
proteins 166201
O feed 17083
oil: food 18395
peak in (2011) xvii vegetable proteins (oilseeds)
and petroleum, history and 196201
fundamentals 75110; see and genetic modification 166,
also oil and petroleum 240
products grains and oilseeds 206
oil and petroleum products 75110, and rotation 2014
77, 78, 80, 81, 82, 84, 85, 86, storage 198
87, 88, 90, 91, 92, 93, 94, 96, trends and swing factors for
97, 100, 102, 104, 107, 108, future 2045
109 Organization of Petroleum
and critical fuel and elasticity Exporting Countries (OPEC)
758 1016 passim, 102
and crude grades and locations modern oil pricing begins with
7983 birth of 101
and crude markets and trading Ospraie Management LLC 385
97101
449
P S
PCA, see principal component Saddam Hussein 104
analysis Saudi Arabia, as swing oil
petroleum and oil products, see oil producer 105
and petroleum products scrap metals 138
PetroPlus 84, 100 see also metals
Ponca 91 sentiment:
precious metals 1457 and commodities 920, 10, 11, 12,
exchanges that list 146 13, 14, 15, 16, 17, 18, 19; see
gold among most actively traded also commodity markets
146 see also SG sentiment indicator
gold market an outlier among 6 SG sentiment indicator, and
many trading centres for 146 commodities 911
most actively traded 146 shale 3642, 38
principal component analysis Shanghai Futures Exchange
(PCA), explained 4 (SHFE) 144, 146, 414, 416,
Purchasing Managers Index(PMI)4 417, 418, 423, 424, 426
Shanghai Securities Exchange 418
R Shenzhen Securities Exchange 418
regasification 54, 56, 58, 603 SHFE, see Shanghai Futures
passim, 61 Exchange
Regulation on Energy Market Shuanghui International 205
Integrity and Transparency Singapore Mercantile Exchange 424
(REMIT) 1301 Smithfield Foods:
Renewable Fuel Standard (RFS) JBS acquires beef business of 200
(US) 253 Shuanghui International
rice, as percentage of worlds acquires 205
dietary energy 192 South Korea, MMT growth of 56
risk-off versus risk-on Soviet era, and assertion for energy
environments 21, 22, 23 dominance 103; see also oil
Russia: and petroleum products:
becomes largest producer of historical price perspective
crude 101 on
growing economic prominence spark and dark spreads 121
of 86 Spot Energy Index 33764
and LNG capacity 63 see also energy indexes: tracking
and natural-gas pipeline exports, storage:
decline in 63 forward curves and market
see also BRIC economies value of 30911, 310
450
INDEX
of grains and oilseeds 198 clean-up and rectification, first

of natural gas 436, 45, 47, 49, 52 phase of 41012
of oil and petroleum 8990 clean-up and rectification,
structural alpha strategies 30535, second phase of 41214
310, 311, 314, 316, 319, 320, companies 4256
323, 324 and corporatisation of
and backtesting bias 3212 exchanges 430
curve placement 308, 30915, development of 4267
311, 314 and exchanges,
market segmentation across corporatisation of 430
commodity forward curves and financial futures,
31112 successful launch of 418
risks of 31315 first ten years (19902000)
seasonality in 31213 410
momentum 3089, 31517 and foreign capital 417
and commodities 32535 further opening up of 431
and commodity beta 326 and futures and spot, and
and downside protection price volatility between
3345 421
and long and short 327, 335 and hedging positions,
and roll yield and excess declaring 422
return 330, 330 and Hong Kong branches of
and sources of excess return future commission
3279 merchants (FCMs) 417
volatility 309, 31820 huge development space of
weaving an alpha basket 3224 428
and insurance function and
T development of options
tail-risk management principles 381 4323
Taiwan, MMT growth of 56 major developing trends
Texas Railroad Commission 101, 42931
103 market operation, micro-
Thailand Futures Exchange 425 characteristics of 420
Three Cheers for US$5 Oil 104 and open interest 4201
trading volume analysis: and opening of futures
and Chinas futures market companies 417
40937, 411, 413, 415, 416, possible changes to 4289
418, 419, 420, 421, 422, 424, and price changes and price
427, 432 hit limits 422
451
and product-listing control, West Texas Intermediate (WTI) 20,

loosening 42930 76
and promotion of new futures wholesale power markets 11331,
contracts 41617 123, 125
regulatory regime, and coming decade, key issues
determining 41415, 415 for 12830
second ten years (200010) changing consumer
41518 requirements 130
and short-term trading activity evolving market design 1289
422 financial market regulation
and stable trading and 12930
stability 4312 electricity 11418
trading characteristics 2011, dispatch arrangements
analysis of 4225 11418
trading volumes (200010) and functioning wholesale
41820 markets, location of 118
trading volumes (2012) 426 locational issues concerning
11617
U market design 11718
United Nations Food and and quality 117
Agriculture Organization and scientific laws 11418
229 and hedging strategies and price
US National Weather Service 66, 67 formation 1212
and historical price perspective
W 1228
weather: and European markets, wider
and agriculture markets, trading relationship between 1267
in 26873; see also Germany 1224, 123
agriculture Great Britain 1246, 125
and commodities 6574 new developments 1278
data basics 667 sources of information
day in the life 6770 concerning 1301
impacts: on agriculture 723 trading of, European markets
impacts: on livestock 734 11821
impacts: on softs 74 design principles, and
impacts: on transport 73 importance of balancing
tropical conditions 702 regime 11819
and North American gas market forward markets 1201
42 Wood River Refinery 91
452
INDEX
Z Zhengzhou Commodity Exchange

ZCE, see Zhengzhou Commodity (ZCE) 414, 416, 417, 418, 423,
Exchange 424, 426
453

Ganapathy Vidyamurthy Pairs Trading

Uploaded by

Copyright:

Available Formats

Ganapathy Vidyamurthy Pairs Trading

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ganapathy Vidyamurthy Pairs Trading

Uploaded by

Copyright:

Available Formats

What are the main commodities discussed in the book?

What are the main commodities discussed in the book?

What are some of the major commodity exchanges mentioned?

What are some of the major commodity exchanges mentioned?

Commodity

Commodity Investing and Trading

Edited by Stinson Gibner

Chapters focus on the fundamentals

This book has been

Commodity Investing and Trading.indd 1 01/10/2013 15:36

Commodity Investing and Trading

Commodity Investing and Trading

Published by Risk Books, a Division of Incisive Media Investments Ltd

2013 Incisive Media

ISBN 978 1 906348 84 7

British Library Cataloguing in Publication Data

Publisher: Nick Carver

Designer: Lisa Ling

Typeset by Mark Heslington Ltd, Scarborough, North Yorkshire

About the Editors ix

PART I: COMMODITY MARKET FUNDAMENTALS

1 The Impact of Non-fundamental Information on Commodity

2 The North American Natural Gas Market 25

3 A Day in the Life of Commodity Weather 65

4 Oil and Petroleum Products: History and Fundamentals 75

5 Wholesale Power Markets 113

6 The Metals Markets 133

COMMODITY INVESTING AND TRADING

7 Grains and Oilseeds 165

PART II: TRADING AND INVESTMENT STRATEGIES

9 Farmland as an Investment 229

10 Agriculture Trading 249

11 Quantitative Approaches to Capturing Commodity Risk

12 Structural Alpha Strategies 307

13 Energy Index Tracking 337

PART III: MARKET DEVELOPMENTS AND RISK MANAGEMENT

14 Enterprise Risk Management for Energy and Commodity

15 Credit Valuation Adjustment (CVA) for Energy and

16 The Past, Present and Future of Chinas Futures Market:

About the Editor

Stinson Gibner is an analyst at Whiteside Energy, having worked in

About the Authors

Kostas Andriosopoulos is executive director of the Research Centre

Francisco G. Blanch is managing director and head of global

Carlos Blanco is managing director of NQuantX LLC, and director of

Greyson S. Colvin is founder and managing partner of Colvin & Co,

COMMODITY INVESTING AND TRADING

numerous publications and is co-author of the Investors Guide to

Rita DEcclesia is a professor at Sapienza University of Rome and

Jay Gottlieb led development of the first coal derivatives instru-

Todd Gross is chief investment officer, managing member and

ABOUT THE AUTHORS

Michael Haigh is managing director and global head of commodi-

Mark Hooker was most recently senior managing director of State

Paul D. Kaplan is director of research for Morningstar Canada and a

Paul R. Lucek is the chief investment officer, Hedge Fund Group,

COMMODITY INVESTING AND TRADING

University, College of Physicians and Surgeons, where as a

Jose Marquez is a meteorologist for Whiteside Energy. Since 2000,