Академический Документы
Профессиональный Документы
Культура Документы
Inflation
Sensitive
Assets
Instruments and Strategies
Edited by Stefania Perrucci
and Brice Benaben
PEFC Certified
This book has been
produced entirely from
sustainable papers that
are accredited as PEFC
compliant.
www.pefc.org
Inflation_Sensitive_Assets.indd 1
12/07/2012 11:18
i
perrucci 2012/7/11 17:12 page i #1
Inflation-Sensitive Assets
i
i
i
perrucci 2012/7/11 17:12 page ii #2
i
i
i
perrucci 2012/7/11 17:12 page iii #3
Inflation-Sensitive Assets
Instruments and Strategies
i
i
i
perrucci 2012/7/11 17:12 page iv #4
i
i
i
perrucci 2012/7/11 17:12 page v #5
i
i
i
perrucci 2012/7/11 17:12 page vi #6
i
i
i
perrucci 2012/7/11 17:12 page vii #7
Contents
About the Editors
ix
xi
Foreword
Mihir Worah
PIMCO
xix
Acknowledgements
xxi
PART I
Inflation-Sensitive Assets
Stefania A. Perrucci
New Sky Capital
1
3
13
25
43
69
79
Inflation-Linked Markets
Gang Hu; Stefania Perrucci
Credit Suisse; New Sky Capital
103
137
vii
i
i
i
perrucci 2012/7/11 17:12 page viii #8
INFLATION-SENSITIVE ASSETS
PART II
9
177
179
209
255
277
299
325
PART III
351
369
389
435
479
Index
507
viii
i
i
i
perrucci 2012/7/11 17:12 page ix #9
i
i
i
perrucci 2012/7/11 17:12 page x #10
i
i
i
perrucci 2012/7/11 17:12 page xi #11
i
i
i
perrucci 2012/7/11 17:12 page xii #12
INFLATION-SENSITIVE ASSETS
for setting equity investment guidelines for the private bank. Steven
also served as a member of the special situations equity strategy
group, and in a variety of new product development projects. He
has a BA from Hunter College and gained a CFA in 1989.
Frank Browne is senior advisor to the Governor, Central Bank of Ireland. He previously worked as senior economist at the OECD (1988
92), as deputy head of Stage 3 (EMU) Division at the European Monetary Institute (19948) and as advisor to the research directorate at
the European Central Bank (19982000). He has been a member of
the ECB Monetary Policy Committee and the ECB Bank Supervision
Committee. Franks research has been published in European Economic Review, Review of Economics and Statistics and The Manchester
School.
Brad Case is senior vice president of the research and industry information group for the National Association of Real Estate Investment
Trusts (NAREIT). His research has been published in Review of Economics and Statistics, Real Estate Economics, the Journal of Real Estate
Finance and Economics, Journal of Portfolio Management and other academic and industry publications. He is the co-inventor of PureProperty indexes of commercial property values as well as backward
forward trading contracts. Brad earned his BA at Williams College,
his MPP at the University of California at Berkeley, and his PhD in
economics at Yale University.
David Cronin is senior economist at the Central Bank of Ireland.
He has represented the Bank at various European System of Central
Banks committees and at other forums. He has published articles in
journals such as Journal of Economics and Business, Empirica and The
Cato Journal, among others.
Ric Deverell is a managing director and head of the commodities research team at Credit Suisse. He previously spent 10 years
at the Reserve Bank of Australia, holding several senior positions
including deputy head of economic analysis and chief manager of
international markets and relations. During the 1990s, Ric worked in
the Department of the Prime Minister and the Australian Treasury,
where he was head of the Asian section during the Asia crisis. His
research interests include analysis of the global business cycle, the
xii
i
i
i
perrucci 2012/7/11 17:12 page xiii #13
emergence of the Asian economies and developments in commodity markets. Ric is a graduate in economics from the University of
Tasmania.
Dennis Eagar manages global listed infrastructure investment portfolios for Magellan Asset Management on behalf of retail and wholesale investors. The team is regarded as expert in the investment
analysis and valuation of infrastructure assets and has more than
50 years collective experience in infrastructure investment. The
team has advised on investment in airports, ports, toll roads, communications infrastructure, pipelines, electricity distribution and
transmission, and water and waste water treatment and distribution.
Robert J. (Bob) Greer is an executive vice president and manager
of real return products at the Pacific Investment Management Company (PIMCO). He managed commodity index business for Daiwa
Securities, Chase Manhattan Bank and JP Morgan. Bob developed
one of the two common methods of explaining sources of commodity
index returns and has spoken on this asset class in college lectures,
on national television, and at industry conferences and trade meetings. He has published in The Journal of Portfolio Management and The
Journal of Derivatives, among others. He has consulted on commodities for the CIA, the Bank of England and the New York Fed. Bob is
the author of The Handbook of Inflation Hedging Investments. He has
a Bachelors degree in mathematics and economics from Southern
Methodist University and an MBA from Stanford Graduate School
of Business.
Ken Griffin is a managing director at Conning, where he heads
the life and health advisory team that is responsible for providing
assetliability and integrated risk management advisory services to
life and health insurance company clients. He also serves as a lead
consultant for various capital management and investment-related
projects involving global life and property and casualty insurance
companies. Prior to joining Conning in 2001, Ken held various positions within Swiss Re Investors ALM unit. He has been involved in
product development, insurance securitisations, ALM and investment portfolio management for insurance companies since 1997. He
is a graduate of the University of North Carolina at Chapel Hill with
a BA in Economics.
xiii
i
i
i
perrucci 2012/7/11 17:12 page xiv #14
INFLATION-SENSITIVE ASSETS
Olesya V. Grishchenko is an economist in the monetary affairs division of the Board of Governors of the Federal Reserve System, conducting research and policy work related to the term structure of
inflation expectations, inflation uncertainty, deflation probabilities
and on the term structure models of real interest rates. She has a
PhD in finance from the Stern School of Business of New York University, and previously worked as an assistant professor of finance
in the Smeal College of Business, Penn State University. Olesya
is a visiting assistant professor of finance at the New Economic
School of Moscow, with research interests in empirical asset pricing,
consumption-based modelling and computational methods. She has
published in the Journal of Economics and Business and Journal of Business and Economic Statistics and is a member of the American Finance
Association and the European Finance Association.
Gang Hu is a managing director of Credit Suisse in the fixed income
division, based in London. Within the fixed-income division he is
the global head of inflation, responsible for the US, UK, EU and
developed Asia in linear and non-linear inflation business. He joined
Credit Suisse in July 2011 from PIMCO, where he was an executive vice president in the real return team, co-managing the largest
inflation fund group in the world. Prior to that, Gang worked at
Deutsche Bank, running the US inflation business. He has 11 years
experience in the banking industry and holds a BA in applied mathematics from Tsinghua University and a PhD in applied mathematics
from California Institute of Technology.
Jing-Zhi (Jay) Huang is a McKinley Professor of business and associate professor of finance at the Smeal College of Business, Penn
State University. His research interests include derivatives markets, credit risk, fixed-income markets, mutual funds and hedge
funds. His papers have been published in The Journal of Finance,
Economic Theory, Journal of Derivatives, Journal of Fixed Income, Journal
of Real Estate Finance and Economics and Review of Derivatives Research,
among others. He won the Best Paper Awards at the Financial Management Association and the Eastern Finance Association Meetings,
and NYUs Stern School Club 6 Teaching Award. He received his
PhD in physics from Auburn University, and his PhD in finance
from New York University.
xiv
i
i
i
perrucci 2012/7/11 17:12 page xv #15
Nicholas Johnson is an executive vice president and portfolio manager at the Pacific Investment Management Company (PIMCO),
where he focuses on commodities and inflation. He previously
managed the portfolio analysts group. Prior to joining PIMCO in
2004, he worked at NASAs Jet Propulsion Laboratory, developing
Mars missions and new methods of autonomous navigation. He has
eight years of investment experience and holds a masters degree
in financial mathematics from the University of Chicago and an
undergraduate degree from California Polytechnic State University.
Kamal Naqvi is a managing director of Credit Suisse in the investment banking division, based in London, where he is the head of
institutional commodity sales. He has over 15 years experience in
the resources industry. He joined Credit Suisse in July 2007 after four
years at Barclays Capital, where he was responsible for coverage of
hedge funds and institutional clients across commodity products,
after previously being a director in the commodity research team.
Prior to that, he was a commodities analyst with Macquarie Bank
in London, after having worked for CRU International in London
as part of both the lead/zinc and precious metals teams. Kamal
began his career as a project manager/economist for the mining
and mineral processing division of the Tasmanian State Government
Department of Development and Resources. He holds degrees (with
honours) in law and in economics from the University of Tasmania.
Jeffrey Oxman is an assistant professor of finance in the Opus College of Business at the University of St Thomas in Minneapolis,
MN, USA. His research has been published in the Review of Derivatives Research and Economics Letters. His research interests include
inflation, value investing and corporate finance. He holds a PhD in
finance from Syracuse University in New York.
Shaun K. Roache is an economist in the Western Hemisphere department of the IMF. He has worked in various countries, including
Brazil and China, contributing to the IMFs analysis of global commodity markets, and worked as part of the team managing the IMFs
investments. Previously, Shaun worked for 10 years as a global financial market strategist, for companies including ING Barings and
Citigroup. He holds a PhD in economics from Birkbeck College,
University of London and is a CFA charterholder.
xv
i
i
i
perrucci 2012/7/11 17:12 page xvi #16
INFLATION-SENSITIVE ASSETS
Gerald Stack manages global listed infrastructure investment portfolios for Magellan Asset Management on behalf of retail and wholesale investors. The team is regarded as expert in the investment
analysis and valuation of infrastructure assets and has more than
50 years collective experience in infrastructure investment. The
team has advised on investment in airports, ports, toll roads, communications infrastructure, pipelines, electricity distribution and
transmission, and water and waste water treatment and distribution.
John A. Tatom is director of research at Networks Financial Institute and associate professor of finance at Indiana State University.
From 2000 to 2005, he was an adjunct professor in the economics
department at DePaul University in Chicago, and during 20034 he
was also a Senior Fellow at the Tax Foundation in Washington, DC.
From 1995 to 2000, John was with UBS in Zurich in various positions,
including executive director and head of country research and limit
control, and chief economist for emerging market and developing
countries. From 1976 until 1995, he was a research official and policy
adviser at the Federal Reserve Bank of St Louis. He has taught at
several colleges and universities and holds a PhD from Texas A&M
University.
Franck Triolaire is an executive director at Morgan Stanley. He has
covered global inflation products since 1999 and is the head of inflation trading for Europe and Asia. He started on the buy-side in
Sinopia, trading inflation products and derivatives. Franck joined
BNP Paribas in 2004 as an inflation trader, took over the euro flow
franchise in 2005 and put BNP Paribas on the top of the rankings
with issuers and customers. He is highly connected in the inflation
space and has managed several government inflation linked bond
issues for France, Germany and Italy. He became co-head of inflation
trading for Europe and Asia in 2008 and head in 2010, promoting
inflation products on cash, derivatives, volatility and structures.
David Vavra runs a consultancy specialising in macroeconomic
modelling and forecasting, helping the pricing of financial instruments in currencies with shallow local markets. He also works as
consultant for the IMF, doing research and applied work on monetary and other issues in a number of countries. David assisted
the National Bank of Serbia in implementing its inflation-targeting
xvi
i
i
i
perrucci 2012/7/11 17:12 page xvii #17
regime and helped to develop forecast-based decision-making systems in the central banks of Colombia, the Czech Republic, Croatia,
Turkey and Peru. He was associated with the Czech National Bank,
leading the macroeconomic modelling and forecasting division and
advising the Governor.
Susan M. Wachter is the Richard B. Worley Professor of financial
management, professor of real estate and Finance at The Wharton
School, University of Pennsylvania. She is also the co-director of the
Penn Institute for Urban Research. From 1998 to 2001 she served
as Assistant Secretary for Policy Development and Research at the
US Department of Housing and Urban Development. She was the
editor of Real Estate Economics from 1997 to 1999 and serves on the
editorial boards for several real estate journals. She is the author of
more than 100 scholarly publications.
Kris Webster manages global listed infrastructure investment portfolios for Magellan Asset Management on behalf of retail and wholesale investors. The team is regarded as expert in the investment
analysis and valuation of infrastructure assets and has more than
50 years collective experience in infrastructure investment. The
team has advised on investment in airports, ports, toll roads, communications infrastructure, pipelines, electricity distribution and
transmission, and water and waste water treatment and distribution.
Edward Y. Yao is a vice president at Conning, where he is responsible
for providing advisory services to property and casualty insurance
company clients with respect to strategic asset allocation and integrated assetliability risk management. Prior to joining Conning in
2008, he was a pricing actuary with Travelers Insurance Co. Edward
has provided actuarial consulting services to a variety of clients since
2000. He holds a degree in economics from Qingdao University,
China, and masters degrees in economics and computer science
from Vanderbilt University.
xvii
i
i
i
perrucci 2012/7/11 17:12 page xviii #18
i
i
i
perrucci 2012/7/11 17:12 page xix #19
Foreword
Inflation ranks among the most insidious and destructive of economic forces.
At the extreme, hyperinflation as seen in Germany during the
Weimar Republic in the 1920s, when prices doubled every few days,
or Hungary in 1946, when prices doubled every 15 hours can
destroy the value of paper money and social cohesion. Even moderate inflation can pick the pockets of savers, erode the competitiveness of industries and nations and wreak havoc with investment
portfolios designed to preserve purchasing power.
Since Paul Volcker broke the back of the Great Inflation in the early
1980s, the US has enjoyed three decades of uninterrupted, low inflation. In fact, in the aftermath of the financial crisis of 2008, deflation,
not inflation, has seemed the larger risk.
Nonetheless, the outlook for inflation has grown more uncertain,
the investment challenges and opportunities more immense. A mix
of unconventional monetary policy and massive fiscal stimulus has
saddled the US with an unsustainable debt load, leaving few solutions: faster economic growth, higher taxes, reduced spending or
inflation. Among these, inflation is the easiest, at least in political
terms, and the most likely: no act of Congress is required.
Moreover, if the Federal Reserve wants to engineer inflation to
lower the nations debt-to-GDP ratio, as it did in the years after
World War II, it will have plenty of help from secular trends. In
developing nations, which over the past 20 years have exerted disinflationary pressure through exports of low-cost goods and services,
nearly 2 billion people are likely to join the middle class over the next
two decades moving from huts to concrete apartments, mopeds to
sedans and putting upward pressure on prices for food, crude
oil, copper and other essential commodities. A slowdown in productivity improvement in emerging economies also could translate
into higher production costs and, ultimately, higher export prices.
Although aggressive inflation of the sort seen in the 1970s appears
unlikely, investors should remain vigilant: history shows that price
spikes can come abruptly, with little warning (think of the Arab oil
xix
i
i
i
perrucci 2012/7/11 17:12 page xx #20
INFLATION-SENSITIVE ASSETS
embargoes of the 1970s). Indeed, its structural roots are so deep that
inflation can make us feel as helpless as individuals against the tide.
Yet it need not be so. Armed with an understanding of the factors
that drive inflation, and the way inflation-sensitive asset classes perform and correlate with each other, it is possible to prepare for and
manage a variety of inflationary scenarios. Inflation does not always
need to be pernicious; it can also be propitious. What is needed is a
dynamic portfolio of inflation-sensitive assets.
This book provides an essential resource for investors, consultants
and service providers keen to preserve wealth. Essays from a wide
range of experts and thought leaders, including some of my colleagues at PIMCO, explore the risk factors that drive inflation and
the tools and vehicles investors can use to realise attractive potential
returns in a variety of inflation scenarios.
Part I examines the building blocks, and related derivatives, of
what should be viewed as a new asset class of inflation-sensitive
assets, including investable commodity indexes, real estate, infrastructure and inflation-linked bonds. Part II moves to a more theoretical level, exploring: models used to forecast inflation and term
structures of interest rates; the interaction of monetary policy, inflation and commodity prices; theories of inflation and its impact on
equities, fixed income and other asset classes; and ways to hedge
inflation through asset and sector rotation. Part III provides practical guidance to investors, with chapters on: strategies for liabilityand real-asset management for insurance, pension and ultra-high
net-worth portfolios; trading tactics; and inflation-linked markets in
emerging countries.
I hope these insights give you new ways to think about inflation
and practical tools for preserving your purchasing power.
Mihir Worah
Managing Director,
Head of Real Return Portfolio Management,
PIMCO
May 2012
xx
i
i
i
perrucci 2012/7/11 17:12 page xxi #21
Acknowledgements
Many people had an important role in the successful completion of
this volume. First in line, our excellent contributing authors: thanks
for your dedication, your hard work and for making this project the
great book it has become.
Over the past year, the team at Risk Books has provided much
needed guidance, enthusiasm and support. It has been a pleasure to
work with them.
Last but not least, to the team at New Sky Capital, and to our
families and friends, thanks for your patience and support during
the creation of this volume.
Stefania A. Perrucci
Brice Bnaben
Philadelphia, June 2012
xxi
i
i
i
perrucci 2012/7/11 17:12 page xxii #22
i
i
i
perrucci 2012/7/11 17:12 page 1 #23
Part I
Introduction:
Markets and Instruments
i
i
i
perrucci 2012/7/11 17:12 page 2 #24
i
i
i
perrucci 2012/7/11 17:12 page 3 #25
Inflation-Sensitive Assets
Stefania A. Perrucci
New Sky Capital
i
i
i
perrucci 2012/7/11 17:12 page 4 #26
INFLATION-SENSITIVE ASSETS
25
20
15
Great Moderation
%
10
5
2010
2005
2000
1995
1990
1985
1980
1975
1970
1965
1960
0
5
bull bond market, and the generally successful anchoring of inflation within a small range. After the Great Moderation in the 1980s
onwards, and an orderly downward convergence of global inflation
rates (Figure 1.1), we are seeing a divergence between developed versus emerging markets, with both deflationary and inflationary forces
at play. As a result of this dichotomy, the necessity to understand and
manage inflation risk is greater than ever.
Policymakers, in particular, will have to design and implement
the right exit strategies, in order to moderate easy monetary stances
in developed countries without choking growth or to control overheating emerging economies without triggering a collapse in asset
prices. It is a delicate balance, as even small shocks to money supply or real growth can have a large effect on medium-term inflation, while a loss of confidence in central banks and policymakers
can undermine the long-term anchoring of inflation expectations.
Because of these inherent sensitivities, there is little room for error.
Of course, inflation represents not just a risk but also an investment opportunity. We shall expand on this in greater detail in
Chapter 19.
4
i
i
i
perrucci 2012/7/11 17:12 page 5 #27
INFLATION-SENSITIVE ASSETS
Other services
20%
Food
15%
Core goods
24%
Shelter
31%
Energy
9%
Values have been rounded. Source: Bureau of Labor Statistics, New Sky Capital.
i
i
i
perrucci 2012/7/11 17:12 page 6 #28
INFLATION-SENSITIVE ASSETS
Bank
US government
pays for goods
US Govt
Treasury
bond
Federal
Reserve
Bank
loans
91
i
i
i
perrucci 2012/7/11 17:12 page 7 #29
INFLATION-SENSITIVE ASSETS
Inflation
Unemployment
Source: New Sky Capital.
and financial assets (real and nominal rates, liquidity spreads, credit
spreads and equities, commodities, real estate, volatility, foreign
exchange rates, etc).
One of the most well-known macro models of inflation is the
Phillips curve (Phillips 1958). This describes the inverse relationship
between inflation and unemployment observed in many countries,
for example, in the UK and the US. Among other things, this relation poses a difficult trade-off for policymakers, who need to balance
price stability with full employment. However, models based on the
Phillips curve were questioned in the 1970s, when many countries,
including the US, experienced stagflation, ie, high inflation in the
midst of stagnant growth. If, at times, inflation and employment do
not move along the Phillips curve, the obvious solution is to move
the curve itself.
Accordingly, new models came about where the relation between
inflation and unemployment (or alternative measures of real economic activity, eg, capacity utilisation) became a dynamical feature.
In these models, inflation and unemployment could move along
the usual Phillips/demand-pull curve, while the curve itself could
be subjected to up or down moves in response to cost shocks (eg,
commodities price spikes) or change in inflation expectations (as
observable in inflation surveys or the inflation linked market). Figure 1.4 shows a schematic representation of expectation-augmented
Phillips curve models.
7
i
i
i
perrucci 2012/7/11 17:12 page 8 #30
INFLATION-SENSITIVE ASSETS
tion),
a change in inflation expectations.
i
i
i
perrucci 2012/7/11 17:12 page 9 #31
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 10 #32
INFLATION-SENSITIVE ASSETS
trend, with several pension funds, insurance funds, and even wealth
funds, changing their traditional asset allocation to include a new
inflation-sensitive sector, along with the more traditional equity,
fixed income and credit sectors. The newly defined asset class has
steadily gained weight in portfolio allocations from several institutional investors. The shift is of monumental importance in the asset
management industry, as these institutions control trillions of US
dollars of capital in the space.
There are tangible benefits in defining inflation-sensitive assets
to include other instruments in addition to inflation-linked cash and
derivative securities.
Diversification: these instruments have distinct return distri-
bution characteristics and macroeconomic drivers, thus providing diversification in real return space. For example, several analytical studies show how portfolios should allocate
to inflation-linked securities at low/moderate return targets,
while other inflation-sensitive assets such as commodities or
equities gain a higher weight as the riskreturn target of the
portfolio is increased along the efficient frontier. Notably, studies made in real (rather than nominal) return space reach even
stronger conclusions in regard to this point (Perrucci 2011b).
Pricing: often a similar macro/inflation view is priced across
i
i
i
perrucci 2012/7/11 17:12 page 11 #33
INFLATION-SENSITIVE ASSETS
See http://www.bls.gov/.
11
i
i
i
perrucci 2012/7/11 17:12 page 12 #34
INFLATION-SENSITIVE ASSETS
REFERENCES
Perrucci, S., 2010, Inflation-Sensitive Assets: Portfolio Benefits and Opportunities, URL:
http://www.newskycapital.com.
Perrucci, S., 2011a, Inflation, Money and Debt, URL: http://www.newskycapital.com.
Perrucci, S., 2011b, Efficient Frontier and Optimal Portfolios in Real vs Nominal Space,
URL: http://www.newskycapital.com.
Phillips, A. W., 1958, The Relationship between Unemployment and the Rate of Change
of Money Wages in the United Kingdom 18611957, Economica 25(100), pp. 28399.
12
i
i
i
perrucci 2012/7/11 17:12 page 13 #35
Bob Greer
PIMCO
In this chapter, we present a brief history of investable commodity indexes, and discuss the link between inflation and commodity
prices.
The first section covers the 1970s, and the effect of energy and food
price spikes on inflation and traditional asset classes. Commodity
futures are discussed next, as well as the authors work in building
the first investable commodity index. A brief recap of the evolution
in commodity indexing and investment vehicles until the present
day1 follows. We then conclude with a discussion of commodities
as an inflation hedge.
i
i
i
perrucci 2012/7/11 17:12 page 14 #36
INFLATION-SENSITIVE ASSETS
Despite these events highlighting the importance of the link between commodities and inflation, investors did not have an effective way to get exposure to commodity prices in order to hedge
inflation. Some tried to purchase natural resource stocks, but those
investments carried other risks as well, in particular a significant
correlation to the overall equity market, with movements in the latter often overshadowing changes in commodity prices themselves.
Other institutional investors, who had very long investment horizons, tried to purchase real commodity-producing assets, such as
farmland. But, besides being illiquid, these investments were also
exposed to other non-essential risks, on both the operational and
finance sides (weather, government regulations, political risk, etc).
All of this set the perfect stage for the developments that followed.
COMMODITY PRICE INDEXES AND FUTURES
There has been interest in commodity prices, and indexes of those
prices, for a very long time. Until the late 1970s, the available indexes
typically referred to prices of physical commodities. Some of these
early indexes were published by Reuters, the Financial Times and
The Economist, among others. They comprised a broad range of commodities, including some for which there were futures markets, as
well as others that had no futures equivalent. There was also a variety of other indexes of cash prices for specific industries within the
commodity asset class, including livestock, energy products and
mining products. Both Dow Jones and the Commodity Research
Bureau published indexes which used the current or spot month
price from commodity futures markets as a surrogate for cash markets, partly because this information was readily available. But, like
those other early price indexes, those based on futures prices were
also not investable, because they could not be replicated. That is,
they simply reported the spot month price as a surrogate for cash
prices, without accounting for the fact that an investor would have
to roll their exposure from a nearby contract to a distant contract
before expiry, which can affect returns, sometimes dramatically. That
is because those early index calculators only wanted a measure of the
level of cash prices; the concept of actually investing in commodity
futures had not yet been considered.
During the inflation and related shortages of several commodities in the 1970s, the interest in commodities as investments began
14
i
i
i
perrucci 2012/7/11 17:12 page 15 #37
i
i
i
perrucci 2012/7/11 17:12 page 16 #38
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 17 #39
various components,
17
i
i
i
perrucci 2012/7/11 17:12 page 18 #40
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 19 #41
Despite the proliferation of commodity indexes in the 1990s, providers mainly sought institutional investors, who might enter into
over-the counter (OTC) swaps to get exposure to their index of
choice. Then, in 1997, the industry saw the first vehicle by which both
institutional and retail investors could get commodity index exposure, when Oppenheimer Funds launched the Oppenheimer Real
Asset Fund (now renamed the Oppenheimer Commodity Strategy
Total Return Fund), benchmarked to the GSCI.
Meanwhile, research efforts to define the characteristics of commodities as a distinct asset class, and to educate investors, were also
ongoing. Several papers were published in this period (see, for example, Greer 1997 and references therein), highlighting the difference
between traditional asset classes (such as stocks and bonds) and
commodities. The first are capital assets, which generate a stream
of cashflows and can be valued using net present value analysis. In
contrast, commodities, albeit investable, do not generate a stream of
cashflows. Their value derives from the fact that they can be consumed, and value analysis is driven mainly by supply and demand,
including estimates of future supply and demand.
Despite these efforts, commodity index investing remained somewhat limited to investment banks desks, and did not achieve mainstream status in the investment community. However, there are some
exceptions worthy of mention. There were a few early adopters of
commodities index investing, which included the Harvard endowment, the Ontario Teachers Pension Plan and two of the largest
pension funds in the Netherlands, PGGM and ABP.8 The Government Investment Company of Singapore also entered this market in
the late 1990s. However, up to the start of the 21st century, it is estimated that only about US$10 billion in capital (most of which was
institutional money, given the very few retail vehicles available) was
invested in commodity indexes.
COMMODITY INVESTING BECOMES MAINSTREAM
In the first decade of the 21st century, demand for commodity
index investment surged, accompanied by several new investment
vehicles being offered in the market. This was partly due to the
losses that equity investors suffered in 2000, making them eager
to find another area for investment. In addition, the asset class
was slowly becoming better understood, as commodity indexes
19
i
i
i
perrucci 2012/7/11 17:12 page 20 #42
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 21 #43
is designed to provide. In other words, even when these strategies have attractive diversification and hedging characteristics in
historical backtests (as they often do), it is doubtful whether these
results can be relied on going forward unless there is a solid fundamental explanation behind those returns. Consequently, these
long/short commodity indexes are more similar to hedge fund
strategies, seeking to produce absolute returns in the commodity
space.
i
i
i
perrucci 2012/7/11 17:12 page 22 #44
INFLATION-SENSITIVE ASSETS
research studies (see, for example, Cavalieri et al 2010) based on historical data. Furthermore, the fundamental link between commodities and traditional metrics of inflation (for example CPI) suggests
that these benefits should continue in the future, at least as long as
the underlying drivers of higher inflation are rooted in commodity
prices.
Of course, commodity index returns can be affected by many
factors, at times dramatically; this happened in the second half of
2008, when a deep global recession triggered a significant collapse
in prices. This highlights the added value that skilful management
can bring to commodity investing, whether in its own right, or
for inflation-hedging purposes. Clearly, dynamic markets cannot be
optimally managed, and the full diversification and hedging benefits delivered, by simply employing a mostly static index-based
strategy. Nevertheless, indexes remain an effective and transparent
way to gain exposure to commodities, as well as benchmarks of asset
class returns.
CONCLUSIONS
Commodities are key price inputs in the basket of goods and services
used to measure inflation. This relationship was brought into focus
in the 1970s, a decade during which the investment community grew
deeply aware of the negative effects inflation can have on traditional
capital assets, such as bonds and stocks.
This relationship, and the successful introduction of indexing in
equity funds, prompted the author to develop the first investable
commodity index in 1978. Investors interest, however, dwindled for
quite some time, and it was only in 1991 that the first commercially
available commodity index was launched. Over time, others followed, and a few pooled vehicles were also established in the space,
albeit the capital committed to commodity strategies remained quite
limited until the turn of the century. It was only in the next decade
that commodity investing became mainstream, with a considerable
proliferation of new indexes and investment vehicles and a virtuous
cycle of positive performance (which was interrupted, if only briefly,
by the global recession of 2008).
Since inflation remains one of the most debated issues at the time
of writing, a discussion of whether a commodity index strategy
22
i
i
i
perrucci 2012/7/11 17:12 page 23 #45
might provide inflation hedging is a very topical one. Indeed, commodity index investing provides two sources of inflation protection:
the underlying collateral and the futures index itself. These hedging
properties are supported by historical data, and may continue in the
future, at least as long as the underlying drivers of higher inflation
are rooted in commodity prices.
Past performance is not a guarantee or a reliable indicator of future
results. All investments contain risk and may lose value. Commodities contain heightened risk including market, political, regulatory and natural conditions, and may not be suitable for all
investors. This material contains the current opinions of the author
but not necessarily those of PIMCO and such opinions are subject to change without notice. This material has been distributed
for informational purposes only and should not be considered as
investment advice or a recommendation of any particular security,
strategy or investment product. Statements concerning financial
market trends are based on current market conditions, which will
fluctuate. Information contained herein has been obtained from
sources believed to be reliable, but not guaranteed.
1
This was the First Index Investment Trust launched on December 31, 1975, now known as the
Vanguard 500 Index Fund.
This in turn resulted in the passage of Onion Futures Act of 1958 (7 USC Chapter 1, Section 131), banning the trading of futures contracts on onions.
In most modern indexes, the collateral is assumed to be invested in 90-day US Treasury bills,
but the guiding principles are analogous.
Because personal computers had not yet been invented, the gathering of data (often by hand
from microfilmed copies of The Wall Street Journal) and calculation of returns were quite
laborious, especially by modern standards.
PGGM stands for Pensioen V/d Gezondheid, Geest and Maatsch Belangen. ABP is the abbreviation for Stichting Pensioenfonds ABP, the pension fund for employers and employees in
service of the Dutch government and educational sector.
One of the largest mutual funds of this sort is the PIMCO CommodityRealReturn Strategy Fund, which invests mainly in commodity futures exposure and uses US Treasury
Inflation Protected Securities (TIPS), rather than T-bills, as collateral. The fund had about
US$27.6 billion in assets as of September 2011.
10 In some other countries, especially emerging economies, food and energy are an even larger
part of the total index than in the US.
11 In contrast, other factors, such as the price of shelter, may be a larger component of the CPI,
but they are not as obviously felt, or reported on, every day.
23
i
i
i
perrucci 2012/7/11 17:12 page 24 #46
INFLATION-SENSITIVE ASSETS
REFERENCES
24
i
i
i
perrucci 2012/7/11 17:12 page 25 #47
In this chapter, we discuss the link between commodity prices, inflation and real economic growth. We start with a review of the effect
commodity prices have on inflation, comparing and contrasting
developed and emerging markets, and analysing 2008 as a detailed
case study. Next, we discuss how higher commodity prices influence monetary policy, and compare different policy approaches from
three major central banks (the US Federal Reserve, the European
Central Bank and the Peoples Bank of China). A section discussing
the relationship between commodity prices and economic growth
follows; the key mechanisms through which commodities influence
the real economy are introduced, and the effect of growth imbalances
between developed and emerging markets is explained. Finally, we
discuss how commodity investments might be used, either to hedge
higher inflation or to lower growth. A summary section concludes
the chapter.
i
i
i
perrucci 2012/7/11 17:12 page 26 #48
INFLATION-SENSITIVE ASSETS
Table 3.1 Commodity prices and inflation (first half of 2008, annualised)
Food
Energy
Headline
inflation
Wt
Mature
economies
3.7
13.3
0.7
7.7
Emerging
economies
8.1
29.5
3.8
7.7
Contrib. Wt Contrib.
Non-food,
non-energy
Wt
Contrib.
1.4
79.0
1.6
0.9
62.8
3.4
i
i
i
perrucci 2012/7/11 17:12 page 27 #49
i
i
i
perrucci 2012/7/11 17:12 page 28 #50
INFLATION-SENSITIVE ASSETS
50
40
30
20
10
0
10
20
30
40
50
6
5
4
3
2
CCI (CRB) Index
US Core CPI
1988 1991 1994 1997 2000 2003 2006 2009 2012
1
0
i
i
i
perrucci 2012/7/11 17:12 page 29 #51
50
40
30
20
10
0
0
10
20
30
40
1
CCI (CRB) Index
China non-food CPI
3
50
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Sources: CRB, Chinese National Bureau of Statistics, Credit Suisse.
it is also interesting that in the eurozone area, where labour markets generally remain more heavily unionised and regulated, similar conclusions can also be reached. In fact, since 1991, when the
European Union started publishing combined eurozone inflation
data, the correlation between annual changes in the CRB Index and
eurozone core (again excluding food and energy) inflation has been
negative, in this case 0.27%, suggesting that, as in the US, commodity price spikes negatively affect relative prices in other consumer
sectors, rather than permanently increasing overall inflation by raising inflation expectation, wages and other non-commodity prices
(Figure 3.2).
This analysis suggests that, while large movements in commodity
prices can have a significant impact on headline inflation in the short
run, there is little clear evidence of significant sustained (long-term)
impact on either core or non-core inflation in the US and Europe, at
least in recent history. This is in line with Cecchetti and Moessner
(2008), who also conclude that4
in recent years, core inflation has not tended to revert to headline,
which suggests that higher commodity prices have generally not
spawned strong second-round effects.
i
i
i
perrucci 2012/7/11 17:12 page 30 #52
INFLATION-SENSITIVE ASSETS
80
60
40
20
0
20
China***
India**
Brazil
Eurozone
Indonesia*
US
Czech Rep
Chile
60
Poland
40
Korea
Indonesia from 2003, interpolating June 2008 to November 2008; India using
WPI; China using non-food CPI (see also endnote 5).
Sources: CRB, Press Information Bureau of India, Eurostat, US Bureau of Labor
Statistics, NBS, Credit Suisse.
emerging economies, with China and India (Figures 3.33.4) showing a relatively high positive correlation between movements in the
CRB Commodity Index and core inflation since around 2009.5
This suggests that the impact of higher commodity prices on core
inflation is substantially higher in some cases, or, using Cecchettis
(2008) framework, that core inflation has tended to revert to headline
inflation in economies such as China and India. However, note that,
in the case of China, the positive correlation might be overstated,
given that the short sample period is dominated by the global recession of 20089, when most prices moderated substantially. In the
following section, we shall elaborate on these points and analyse
what happened in 2008, and give potential explanations for these
positive correlations.
WHAT HAPPENED IN 2008?
In the 18 months from January 2007 to June 2008, the US dollar price
of oil increased by 135%, while the price of food (as measured by
the United Nations Food and Agriculture World Food Price Index)
increased by 65%. In the 12 months from July 2007 to June 2008, the
average US dollar price of oil increased by 95%, while food prices
increased by 44%. The direct impact of food inflation was greater
than that of energy in both emerging and mature economies. It is
30
i
i
i
perrucci 2012/7/11 17:12 page 31 #53
Figure 3.5 Food and energy contribution and share of CPI (first half of
2008, annualised)
Food
Energy
9
8
7
6
%
5
4
3
2
1
0
Mature
economies
Emerging
economies
Mature
economies
Emerging
economies
striking, however, that food contributed significantly more to inflation in emerging economies than that in mature economies. It is also
notable that despite being the same weight in CPI baskets, energy
prices had a far larger impact on CPI inflation in mature economies
than that in emerging economies.
Figure 3.5 shows clearly that in mature economies 55% of the
increase in inflation in this period was attributable to food and
energy prices (the key commodity drivers), with only 45% arising from core inflation. For emerging markets, 57% of inflation
comes from food and energy, and 43% from core inflation. This
should not be terribly surprising given the far greater weight of food
and energy components in the average CPI basket in the emerging
markets (nearly 40%) compared with about 20% on average across
mature economies (Table 3.1). Notably, while food and energy prices
made a significant contribution to inflation in emerging economies,
the contribution from core inflation was much higher in these
economies. These findings, coupled with the higher correlation
between commodity prices and core inflation, suggest that commodity price inflation is more of an issue for emerging economies than
developed ones.
For emerging (but not for mature) markets, empirical evidence
shows that commodity prices, and food in particular, have had an
impact on core inflation. However, even in those cases, the increase
31
i
i
i
perrucci 2012/7/11 17:12 page 32 #54
INFLATION-SENSITIVE ASSETS
has not been structural in nature (at least since 1998) as higher core
inflation has reverted to previous (lower) levels in line with the
subsequent stabilisation (or fall) in commodity prices.
INFLATION AND MONETARY POLICY
When assessing the likely monetary policy response, a key question
is whether commodity price inflation is mainly driven by supply
(like the oil crisis in the 1970s) or demand shocks. For example, in the
case of food prices, the predominant cause has typically been supply disruptions,6 due to droughts, floods and crop diseases. While
these disruptions can affect food prices for some time, they generally
self-correct in the medium to long run. Instead, in the first decade
of the 21st century, much of the increase in oil prices was driven by
stronger than anticipated demand, although supply shocks (such as
the one experienced in early 2011) remain a real risk, particularly in
the light of political turmoil in producing countries in North Africa
and the Middle East. For basic materials, the predominant cause of
prices trending up (35% for copper and 100% for iron over the first
decade of the 21st century) was stronger than anticipated demand,
driven by China and other countries. Such demand-side fundamentals are likely to remain the main driver of basic metals prices for
the foreseeable future. In summary, commodities price movements
in the first decade of the 21st century have often had very different
drivers from the oil supply shock of the 1970s.
An important question is whether the increase in prices is likely
to be permanent or temporary (such as a one-off price adjustment),
although in practice it is difficult to assess which one of these will be
ex ante. To the extent that changes are driven primarily by temporary
factors (such as most supply shocks), many central banks would be
expected to look through the impact when setting monetary policy. This is partly because monetary policy has very little impact
on factors such as food prices. But it is also because, by the time
monetary policy can reasonably have begun to affect broader prices
(given the long and variable lags), the price change is already likely
to have reversed. The objective of monetary policy is not to control
short-term inflation fluctuations but to make sure that, on average,
inflation remains within an acceptable range. On the other hand,
when changes in commodity prices are driven by a demand shock,
they are likely to prove more resilient, increasing the likelihood that
32
i
i
i
perrucci 2012/7/11 17:12 page 33 #55
policymakers will indeed intervene. Given the difficulties in assessing the effects of all of these factors, several central banks tend to
focus on inflation expectations. If the latter remain well anchored,
there is scope for monetary policy to look through a (likely) temporary period of higher inflation, primarily because firms and consumers are doing the same. This said, and for a variety of reasons
(some related to differences in the specific countrys inflation process, and some related to cultural and historical norms), it is clear
that there are significant differences in how the major central banks
assess inflation and react to a change in commodity prices. For example, at least historically, the US Federal Reserve has not been overly
concerned about the effect of higher commodity prices on headline inflation, as long as core inflation remained well behaved. This
stance is likely to be reaffirmed, given the relatively low level of core
inflation and large output gap (extra capacity in the economy) at the
time of writing. In contrast, the European Central Bank (ECB), and
before that the Deutsche Bundesbank, has had an explicit headline
inflation target. The ECB will tighten monetary policy if headline
inflation moves above its target because of commodity prices, even
if the move is likely to be transitory. The Peoples Bank of China has
generally adopted a flexible and pragmatic approach, using monetary policy in order to ensure that commodity price inflation (particularly food inflation) does not unduly affect core prices and inflation
expectations.
COMMODITIES AND GROWTH
As with inflation, the impact of commodity prices on the macroeconomy, and real growth in particular, has been the subject of much
academic research. However, while there are many articles on the
topic, most of the literature is heavily focused on oil and the impact
on the US economy.7 As Rasmussen and Roitman (2011) point out
in a recent IMF working paper, there has been much less work on
other countries and very little of that on developing economies.
In addition, most of the research focuses exclusively on the impact
of oil.
There are two key mechanisms that are discussed when assessing
the impact of commodities inflation on economic growth. The first
is that an increase in commodity prices can lead to higher inflation
and therefore result in tighter monetary policy than would otherwise
33
i
i
i
perrucci 2012/7/11 17:12 page 34 #56
INFLATION-SENSITIVE ASSETS
have been the case. This would in turn reduce the pace of economic
growth. The second is that higher commodity prices can act as a
tax on consumers and business, lowering profits and reducing consumption and investment. While there is merit in both arguments,
as often in economics, both rely on strong assumptions, which are
not likely to hold consistently over time.
On the first point, while there is no doubt that rapid increases in
commodity prices can result in higher inflation, the policy response,
if any, is likely to vary significantly depending on the stage in the
economic cycle, and each specific countrys approach to monetary
policy. For example, in 2011 the increase in commodity prices proved
far more problematic for emerging economies, where the commodity weight in inflation baskets is larger and where there is little spare
capacity. In contrast, in mature economies such as the US, commodity inflation has been fairly inconsequential for policymakers, in
view of the large output gap and slow economic growth.
On the second point, while there is much focus on higher commodity prices being a tax on consumers, this effect has been most
relevant in developed markets, and in the US in particular. From a
global perspective, movements in commodity prices are by definition a zero-sum game, with some countries (or corporations) benefiting from higher revenues, while others face the opposite side of the
coin. While there will be frictional issues, as the negative effect on
consumers will manifest in less time than is required for the beneficiaries of higher prices (be they corporations or countries) to translate
higher incomes into higher investments and growth, the ultimate
impact at the global level should be one of distribution rather than
creation or destruction of wealth. Therefore, it is a mistake to focus
only on the cost-side impact of higher commodity prices, especially
in an increasingly globalised economy. For example, while higher
commodity prices might indeed impact negatively on consumers
and producers in G7 economies, many of those same consumers
may own shares in multinational commodity companies, and may
stand to benefit directly from higher equity valuations and higher
dividend payments.
As the global economy continues to recover from the Great Recession of 20089, the distributional issues related to higher commodity prices in the macroeconomy are likely to be more pronounced
than normal. In simple terms, many of the countries that benefit
34
i
i
i
perrucci 2012/7/11 17:12 page 35 #57
i
i
i
perrucci 2012/7/11 17:12 page 36 #58
INFLATION-SENSITIVE ASSETS
10
8
Forecast
6
4
2
0
1860
1880
1900
1920
1940
1960
1980
2000
As mentioned before, there are also beneficiaries of higher commodity prices. Some examples are the oil-exporting countries and
Australia. The latter is an interesting case, with higher commodity
prices resulting in a massive surge in mining investment (Figure 3.6),
which is likely to continue. The interesting thing is that this sort of
investment is capital and import intensive, and our ballpark estimate
is that approximately half of the value added will be from imports.
Consequently, while the impact from higher commodity prices is
a tax on consumers in the US, many American companies will be
clear beneficiaries from greater exports, which will support mining
investments in Australia.
The behaviour of oil exporters such as Saudi Arabia will also be
key for the net global impact of higher commodity prices on growth.
In the five years from 2003 to 2008, the massive increase in oil revenues to the Middle East and North Africa (MENA) was primarily reinvested in the US bond market, contributing to the low-rates
conundrum during the period. In this case, higher capital from oil
revenues was not spent on goods and services but invested in US
bonds. In other words, investment of MENA oil revenues in the US
bond market flattened the yield curve and resulted in looser credit
conditions than would otherwise have been the case. Given the significant political unrest in the MENA region in 2011, the more likely
outcome is that the countries from the region will spend a larger
36
i
i
i
perrucci 2012/7/11 17:12 page 37 #59
proportion of oil revenues on building infrastructure and supporting social programmes domestically. In turn, this may still benefit other countries, since imports of both consumer and investment
goods from developed markets into MENA are likely to increase
substantially.
Another key factor when considering the impact of higher commodity prices on the real economy is the underlying cause of the
increase in prices. If commodity prices are rising because of stronger
demand, as is the case for most non-corn agricultural commodities, it
is unlikely for economic growth to slow substantially. If, on the other
hand, higher prices are the result of a supply shock (as occurred in
the 1970s and early in 2011), the impact on growth is likely to be
more substantial.
COMMODITIES INVESTING
As highlighted above, commodity prices are affected by and have an
effect on both inflation and real economic growth. In the following,
we shall briefly explore how commodity investments can be used to
exploit these relationships with inflation and real economic growth.
Commodity investments and inflation
The market for inflation protection has grown markedly. As an example, the inflation-linked bond market has reached a notional size of
over US$1.5 trillion, and other inflation sensitive assets such as property, infrastructure and commodities have all attracted investors
attention specifically due to their inflation-hedging characteristics.
Despite the evidence that it is emerging markets investors that
have the most to gain from an investment in commodities as a
hedge to inflation, it has been developed market investors that have
most commonly cited inflation as the basis for their investment in
commodities. This may be in part a reflection of different investment approaches, with emerging market investors typically having
an absolute return objective, and developed market institutional
clients typically following a more diversified portfolio approach.
For the latter, indexes consisting of commodity futures, such as the
Standard & Poors Goldman Sachs Commodities Index (SPGSCSI),
Dow Jones UBS Commodities Spot Index (DJUBSCI) or Credit Suisse
Commodity Benchmark (CSCB), have become the most commonly
used investment vehicles to gain exposure to commodities. Such
37
i
i
i
perrucci 2012/7/11 17:12 page 38 #60
INFLATION-SENSITIVE ASSETS
indexes use liquid, transparently priced futures contracts and follow a specific weighting and trading methodology. Importantly, they
can be back-tested: for example, the SPGSCI has been tradeable since
1992, although in academic studies researchers have reconstructed
the index going back to 1959 (Chapter 2).
Commodity indexes saw significant evolution in the early 2000s,
and at the time of writing a range of them incorporate features
such as yield optimisation, thematic weighting or dynamic active
elements. These indexes are also highly customisable, conditional
on the liquidity of the underlying commodities, with country or
regional inflation weighting being offered by several banks. As an
example, Credit Suisse has built and traded an index, the Credit
Suisse Commodity Benchmark China Index, of those commodities that are most sensitive to commodity demand from China.
There are, of course, other ways in which commodity exposure can
be obtained. An investor may choose to trade commodity futures
directly, although this requires knowledge of both the fundamentals and technical characteristics of each specific commodity sector.
More commonly, investors (and retail investors in particular) take
exposure to a customised commodity basket, through a structure
providing derivative exposure to specific commodities over a set
period of time (and sometimes capital protection). Although these
structured investments can be tailored to the specific investor, they
tend to be significantly less liquid and lack the transparency of a
plain vanilla commodities index.
There is an active debate about whether to invest in commodityrelated equities or directly in the underlying commodities. This is
not a black-and-white debate, as there will be certain situations
when one or the other will be the better choice. That said, over the
first decade of the 21st century, the rise in extraction and development costs, widespread geopolitical risks, the popularity of resource
nationalism and a raft of natural disasters have demonstrated that
the ability of a resources company to grow profitability consistent
with rising gross revenue from higher commodity prices is debatable. In turn, this has resulted in a shift towards direct commodity
investments.
This shift has been aided by the development of Exchange Traded
Products (ETPs) offering commodity exposure. These vehicles have
allowed a much wider cross-section of investors to gain commodity
38
i
i
i
perrucci 2012/7/11 17:12 page 39 #61
i
i
i
perrucci 2012/7/11 17:12 page 40 #62
INFLATION-SENSITIVE ASSETS
Other, more subtle, factors are also important. For example, in the
US, oil might be a greater challenge to growth (and equity prices),
while in China agricultural prices might pose the greatest risk. Thus,
it is unlikely that any global commodities index will provide the
best solution for all. Rather, the appropriate vehicle (be it a specific
commodities basket or some other commodity derivative structure)
needs to be tailored to the specific needs and views of the investor.
CONCLUSIONS
In this chapter, we have analysed the connection between commodity prices, inflation and economic growth. We discussed the different
factors influencing developed and emerging markets, in particular
the effect of food and energy prices on headline inflation and the
correlation between commodity prices and core inflation, and how
these depend on the underlying cause of commodities inflation, ie,
whether the price change results from a supply or demand shock. We
also discussed the likely monetary response. Finally, we discussed
commodity investments either as a way to hedge shocks in overall
prices, ie, inflation, or for economic growth.
1
See, for example, Cecchetti and Moessner (2008), Fry et al (2009), Hobijn (2008) and Lipskey
(2008).
In this chapter, we generally use the exclusion method when calculating core inflation,
which simply excludes energy and food from the basket of services and goods. However,
this is done for simplicity, given that other statistical measures such as trimmed mean and
weighted median generally give a better feel for the underlying tendency of inflationary
pressures.
The direct first-round impact of commodity price changes on most finished goods is generally
small. For example, for laptop computers and mobile telephones it is a tiny fraction, while for
US-made cars it is less than 10% of final sale price (pension obligations, in contrast, account for
as much as 25%). For residential apartments, the total cost of raw materials (steel, aluminium,
copper, etc) is generally less than 10% of the cost of construction, and an even smaller fraction
of sale price (developer margins are generally greater than raw material costs).
Similar conclusions are drawn for Australia in a 2010 study conducted by Norman and
Richards (2010), who find little evidence that either commodity prices or the growth rate
of money directly influence Australian underlying inflation.
Note that there are some limitations on data. China has published a core measure of inflation
only since 2006, while India does not publish a CPI. We have therefore used the Wholesale
Price Index for India.
While supply disruptions have been the predominant short-term cause of spikes in food prices
in the early 2000s, it is possible that over time increased demand from emerging markets could
slow or even halt the long-term downwards trend in food prices evident at least since the
beginning of the 20th century. In addition, government policies can have an impact. For
example, the introduction of ethanol mandates in the US has directly contributed to the
doubling of global corn consumption growth, from 0.8% per year in the period 19752003, to
1.6% per year since then. This policy is likely to lead to higher food prices over coming years
(all other things being equal).
40
i
i
i
perrucci 2012/7/11 17:12 page 41 #63
See, for example, Hamilton (1983, 1996, 2005, 2009), Bernanke et al (1997) and Blanchard and
Gal (2007).
REFERENCES
Bernanke, B., M. Gertler and M. Watson, 1997, Systematic Monetary Policy and the
Effects of Oil Price Shocks, Brookings Papers on Economic Activity 28(1), pp. 91157.
Blanchard, O., and J. Gal, 2007, The Macroeconomic Effects of Oil Price Shocks: Why
are the 2000s So Different from the 1970s?, NBER Working Paper 13368.
Cecchetti, S., and R. Moessner, 2008, Commodity Prices and Inflation Dynamics, BIS
Quarterly Review, December, pp. 5566.
Fry, R., C. Jones and C. Kent (eds), 2009, Inflation in an Era of Relative Price Shocks,
in Proceedings of Reserve Bank of Australia Conference 2009, Kirribilli, NSW, URL: http://
www.rba.gov.au/.
Gorton, G., and K. Rouwenhorst, 2005, Facts and Fantasies about Commodity Futures,
Working Paper 04-20, Yale Information Center for Finance.
Hamilton, J., 1983, Oil and the Macroeconomy since World War II, Journal of Political
Economy 91(2), pp. 22848.
Hamilton, J., 1996, This Is What Happened to the Oil Price/Macroeconomy Relation,
Journal of Monetary Economics 38(2), pp. 21520.
Hamilton, J., 2005, Oil and the Macroeconomy, in S. Durlaf and L. Blume (eds), The New
Palgrave Dictionary of Economics, Second Edition (London: MacMillan).
Hamilton, J., 2009, The Causes and Consequences of the Oil Shock of 200708, NBER
Working Paper 15002.
Hobijn, B., 2008, Commodity Price Movements and PCE Inflation, Federal Reserve Bank
of New York Current Issues in Economics and Finance 14(8), pp. 17.
Lipskey, J., 2008, Commodity Prices and Global Inflation, Speech to International Monetary Fund, May, URL: http://www.imf.org/external/nap/speeches/2008/050808.htm.
Norman, N., and A. Richards, 2010, Modeling Inflation in Australia, Discussion Paper
RDP 2010-03, Economics Analysis Department, Reserve Bank of Australia, Sydney.
Rasmussen, T., and A. Roitman, 2011, Oil Shocks in a Global Perspective: Are They Really
That Bad? Working Paper WP/11/194. International Monetary Fund.
Soss, N., and J. Feldman, 2011, Commodities and Core CPI: As Dead as Disco, US
Economics Digest, February 17.
Stevens, G., 2008, Commodity Prices and Macroeconomic Policy: An Australian Perspective, Reserve Bank of Australia Bulletin, June, URL: http://www.rba.gov.au/.
41
i
i
i
perrucci 2012/7/11 17:12 page 42 #64
i
i
i
perrucci 2012/7/11 17:12 page 43 #65
In this chapter we analyse the inflation sensitivity of real estate investments, comparing them to other inflation-sensitive assets. The
most transparent source of real estate investment returns comes
from publicly traded stocks of real estate investment trusts (REITs).
We examine the available return data, with an emphasis on their
relationship to US inflation, although our conclusions may apply
elsewhere as well.
Consumer price inflation (CPI) in the US was 13.5% during 1979,
the worst year since 1947. Dividend income from REITs traded
through the stock exchange averaged 21.2% that year, and total
returns amounted to 24.4%, not only preserving but increasing for
REIT investors the purchasing power that they had lost to inflation.
Inflation averaged 11.6% per year during 197880, the worst threeyear period in six decades; again, however, publicly traded equity
REITs outpaced inflation, with income and total returns averaging
12.2% and 23.1% per year, respectively. The period 197481 was the
most inflationary eight years in the history of the Consumer Price
Index at 9.3% per year, but equity REIT returns easily preserved purchasing power, with income and total returns averaging 10.2% and
16.3% per year.
During the first eight months of 2011, annualised consumer price
inflation was 5.1%. Again, equity REIT returns protected purchasing power, with annualised total returns averaging 8.4%. However,
dividend income, during a period of extraordinary weakness in real
estate operating fundamentals, fell short of inflation at 3.4% per year.
43
i
i
i
perrucci 2012/7/11 17:12 page 44 #66
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 45 #67
i
i
i
perrucci 2012/7/11 17:12 page 46 #68
INFLATION-SENSITIVE ASSETS
direction. For example, as discussed below, energy-related supplyshock inflation episodes such as the oil shocks of the 1970s will affect
real estate returns differently from demand shock inflation deriving
from either monetary or fiscal policy. Moreover, over the real estate
cycle, the inflation-adjusted return of real estate assets will increase
when real estate assets are in demand, and decrease when supply is
plentiful.
In practice, although many REITs have contractually specified
step-up clauses, actual responsiveness of rents, especially to shortrun and intermediate-run effects associated with shocks and cycles,
tends to be dampened by the use of leases.4 As typical lease structures differ by property type, the dampening effect on rent adjustments will also differ: a topic that will be investigated empirically
later in the chapter.
Another conceptual issue that we briefly consider here relates to
the debt structure of real estate investments. It is plausible to assume
that, if an asset were to be financed with long-term fixed-rate debt,
higher inflation would be beneficial to the liabilities of the real estate
owner, as the loss to the debt investor is mirrored as a gain to the borrower. This suggests an empirically testable hypothesis: REITs holding relatively large amounts of long-term fixed-rate debt will tend
to have stronger returns than REITs holding small amounts of longterm fixed-rate debt, during periods of high inflation. Unfortunately,
while data on total debt is readily available through sources such as
SNL Financial,5 data on the composition of debt (long-term versus
short-term, and fixed-rate versus variable-rate) is more difficult to
collect consistently, which is why we have only briefly examined
this here.
As a proxy for detailed information on REIT use of debt, we identified 41 equity REITs that generally used relatively high leverage
and 41 that generally used relatively low leverage, and compared
their returns during months of high versus low inflation during the
period 19912010, where high-inflation months were identified relative to the median monthly CPI over the study period. The empirical
data supports the hypothesis, though not strongly enough to reject
the null: the median monthly total return of high-leverage REITs
(1.26%) exceeded that of low-leverage REITs (1.16%) during months
of higher-than-median inflation, but fell short (1.05% versus 1.76%)
during months of lower inflation.6
46
i
i
i
perrucci 2012/7/11 17:12 page 47 #69
i
i
i
perrucci 2012/7/11 17:12 page 48 #70
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 49 #71
Number of observations
90
83
80
70
60
53
51
50
40
30
20
26
16
18
15
7
10
0
14
<0
12
34
56
78
18
6
910
11+
i
i
i
perrucci 2012/7/11 17:12 page 50 #72
INFLATION-SENSITIVE ASSETS
65.8
70.4
60.8
53.8
43.2
Source: FTSE NAREIT All Equity REITs Index, S&P Goldman Sachs Commodity Index, Ibbotson Associates US TIPS, Barclays Capital US Treasury
TIPS Index, S&P 500 Index, S&P GSCI Gold Index.
Table 4.1 presents the proposed measure of inflation protection using six-month periods constructed from monthly data for a
selection of five asset classes:
1. publicly traded equity REITs, measured by the FTSE NAREIT
All Equity REITs Index;
2. commodities, measured by the S&P Goldman Sachs Commodity Index (GSCI);
3. TIPS as measured by the Ibbotson Associates synthetic US TIPS
series, which is equal to the Barclays Capital US Treasury TIPS
Index from January 1997 onwards, but is backfilled by Ibbotson
prior to that;10
4. US equities, as measured by the S&P 500 Index;
5. gold as measured by the S&P GSCI Gold Index.11
All indexes measure total returns (ie, income plus price appreciation).
As Table 4.1 shows, the two assets providing the most dependable
inflation protection (by our measure) were commodities and equity
REITs, with commodities providing total returns that equalled or
exceeded inflation during 70.4% of high-inflation semesters and
equity REITs close behind at 65.8%. Stocks and TIPS provided somewhat weaker inflation protection by this measure, with stocks protecting purchasing power during 60.8% of high-inflation six-month
periods and TIPS even lower at 53.8%. By our measure, the weakest inflation protection among this group of assets was provided by
gold, which successfully protected purchasing power during only
43.2% of high-inflation six-month periods.12
50
i
i
i
perrucci 2012/7/11 17:12 page 51 #73
i
i
i
perrucci 2012/7/11 17:12 page 52 #74
INFLATION-SENSITIVE ASSETS
65.8
70.4
60.8
53.8 43.2
1 month
51.5
2 months
58.2
12 months
68.9
67th percentile = 4.29
59.4
80th percentile = 4.89
55.0
90th percentile = 8.65
65.0
S&P GSCI Energy Index
S&P GSCI Non-Energy Index
Barclays Capital TIPS Index
55.4
61.7
75.5
70.7
66.3
55.0
75.3
61.0
53.0
56.2
71.4
54.9
50.0
55.0
51.0
50.2
56.6
42.9
41.3
27.5
50.0
44.8
46.4
46.6
52.5
60.0
56.3
Source: FTSE NAREIT All Equity REITs Index, S&P Goldman Sachs Commodity Index, Ibbotson Associates US TIPS, Barclays Capital US Treasury
TIPS Index, S&P 500 Index, S&P GSCI Gold Index, S&P GSCI Energy
Index, S&P GSCI Non-Energy Index.
i
i
i
perrucci 2012/7/11 17:12 page 53 #75
i
i
i
perrucci 2012/7/11 17:12 page 54 #76
INFLATION-SENSITIVE ASSETS
of the 1970s and early 1980s. The analysis shown in Table 4.1 used
the synthetic TIPS return series estimated by Ibbotson Associates,
while row 9 of Table 4.2 tests the robustness of the TIPS results by
reporting the inflation-protection dependability computed using an
alternative synthetic TIPS return series estimated by Barclays Capital. As the numbers show, the results are not very sensitive to the
TIPS return series used, with the Barclays Capital synthetic TIPS
Index returns equalling or exceeding inflation in 56.3% of the 199
high-inflation periods in our sample, comparable to the 53.8% computed using the Ibbotson synthetic TIPS Index. The fact that the
two synthetic indexes show similar results does not, of course, rule
out the possibility that both indexes may be affected by common
methodological biases in backfilling historical returns.
USING TACTICAL PORTFOLIO ALLOCATION FOR INFLATION
PROTECTION
Many investors may choose to shift asset class selections tactically
over time, based on their outlook on inflation and other relevant
variables. The asset classes considered display substantial differences in returns during high- versus low-inflation periods, making
this tactical asset selection option both valuable and risky. In other
words, correct insight on future inflation (and its root causes) can
greatly enhance investment returns, while mistakes in forecasting
can prove costly. During the overlapping six-month periods that we
have identified as high-inflation semesters, the average annualised
rate of inflation was 6.1%, compared to 1.8% during low-inflation
semesters.
As Table 4.3 shows, during high-inflation semesters, commodities
provided by far the strongest average annualised returns at 19.2%
per year. This should not be surprising, as commodities not only
account for a substantial share of the CPI but also have been a driver
of short-term inflation spikes13 in the historical window of our study.
Equity REITs and stocks also provided strong returns, averaging
12.3% per year for equity REITs (with income-only returns averaging
8.6%) and 10.2% per year for stocks. The average return on TIPS
barely beat the inflation rate at 6.9%, while gold fell short of the
inflation rate at 6.0%.
How much the ability to discern between high- and low-inflation
regimes affects the outcome is clear by comparing the previous
54
i
i
i
perrucci 2012/7/11 17:12 page 55 #77
6.1
19.2
12.3
8.6
10.2
6.9
6.0
Source: FTSE NAREIT All Equity REITs Index, S&P Goldman Sachs Commodity Index, Ibbotson Associates US TIPS, Barclays Capital US Treasury
TIPS Index, S&P 500 Index, S&P GSCI Gold Index.
1.8
13.7
6.9
13.0
9.2
7.2
2.4
Source: FTSE NAREIT All Equity REITs Index, S&P Goldman Sachs Commodity Index, Ibbotson Associates US TIPS, Barclays Capital US Treasury
TIPS Index, S&P 500 Index, S&P GSCI Gold Index.
i
i
i
perrucci 2012/7/11 17:12 page 56 #78
INFLATION-SENSITIVE ASSETS
price spikes (for example, oil supply shocks), which are virtually
impossible to predict but whose effects tend to average out over
longer investment horizons.
These results (as shown in Table 4.4), if taken at face value, suggest
that, at least in the historical period considered, the winning allocation strategy would have been to shift the portfolio aggressively into
commodities during periods of high inflation, shifting back to assets
that fulfil other investment goals (eg, income, risk-adjusted returns,
diversification), such as equity REITs or TIPS, during periods of low
inflation. Of course, the difficulty with successfully implementing
such a strategy rests in the ability to predict the inflation regime
during the next several months; the consequences of being wrong
are eloquently showcased by the variance in asset class returns that
characterises different inflation regimes.
Although investors typically focus on the risk of high inflation, low inflation (or even deflation) can be equally insidious for
the returns of a portfolio. While insightful asset allocation by the
active investor has the potential to enhance portfolio returns in
either regime considerably, the objective of a strategic asset allocator or hedger is not to select each periods best performing asset
class but, quite to the contrary, to build effective protection against
both inflation and deflation shocks, thus minimising or eliminating
any reliance on the difficult and risky task of correctly forecasting
inflation going forward.
A BALANCED APPROACH TO THE INFLATION-PROTECTED
PORTFOLIO
The fact that various assets respond differently to inflation suggests
that a blended portfolio of assets with differing inflation-protection
properties may provide a better bulwark against inflation than any
asset in isolation. To illustrate this point, Figure 4.2 summarises the
results of a Markowitz (1952, 1959) meanvariance portfolio optimisation exercise conducted using historically realised real returns
(that is, returns in excess of inflation) for the five inflation-sensitive
assets considered during the 199 high-inflation semesters in our data
sample.
Each point in Figure 4.2 shows an optimal asset allocation, from
the minimum-variance portfolio at the left edge (annualised average real return 3.6%, volatility 9.4%, Sharpe ratio14 0.31) to the
56
i
i
i
perrucci 2012/7/11 17:12 page 57 #79
100
US TIPS
78
80
76
70
74
60
Commodities
50
40 S&P 500
30
20
10
0
3.6
72
70
Gold
90
68
Equity REITs
66
64
4.8 6.1 7.3 8.6 9.8 11.1 12.3 13.6 14.8 16.1
Portfolio expected inflation-adjusted return (%)
100
90
80
US TIPS
70
S&P 500
60
50
40
30
Commodities
20
Equity REITs
10
Gold
0
7.7 8.7 9.7 10.6 11.6 12.5 13.5 14.5 15.4 16.4 17.4
Portfolio expected inflation-adjusted return (%)
i
i
i
perrucci 2012/7/11 17:12 page 58 #80
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 59 #81
Origination
criterion
Period(s) Ann.
Ann.
Success
of
return volatility Sharpe
rate
inflation
(%)
(%)
ratio
(%)
Maximum
success rate
(high-inflation
periods)1
High
Low
All
12.3
8.2
10.2
18.0
23.2
20.9
0.65
0.27
0.43
77.9
Maximum
Sharpe ratio
(all periods)2
Maximum
Sharpe ratio
(high-inflation
periods)3
High
Low
All
High
Low
All
5.6
9.5
7.5
12.8
8.2
10.5
10.5
12.4
11.7
18.8
24.3
21.9
0.48
0.60
0.54
0.65
0.26
0.42
74.9
Maximum
Sharpe ratio
(low-inflation
periods)4
Minimum
variance
(all periods)5
High
Low
All
2.4
9.0
5.7
9.8
8.9
9.9
0.19
0.79
0.45
57.8
High
Low
All
High
Low
All
3.1
7.6
5.3
3.6
8.3
5.9
9.6
8.3
9.3
9.4
9.0
9.5
0.25
0.67
0.44
0.31
0.70
0.49
67.3
High
Low
All
2.0
7.7
4.9
10.0
8.1
9.5
0.14
0.71
0.38
59.8
Equal Sharpe
High
ratios in
Low
high- and lowAll
inflation periods8
6.0
8.3
7.1
10.4
12.0
11.3
0.52
0.52
0.52
75.4
Minimum
variance
(high-inflation
periods)6
Minimum
variance
(low-inflation
periods)7
75.9
69.3
1 Allocation:
55.1% commodities, 39.1% REITs, 4.6% TIPS, 1.2% equities, 3.3% gold. 2 Allocation: 48.9% TIPS, 16.9% REITs, 14.6% commodities, 13.9% equities, 5.8% gold. 3 Allocation: 58.1% commodities, 41.3%
REITs, 0.5% equities, 0% gold. 4 Allocation: 78.7% TIPS, 10.8% equities,
5.4% gold, 5.0% REITs, 0% commodities. 5 Allocation: 80.9% TIPS, 8.4%
commodities, 5.8% gold, 4.9%, 0% REITs. 6 Allocation: 73.3% TIPS, 8.6%
commodities, 8.4% equities, 5.0% gold, 4.8% REITs. 7 Allocation: 90.4%
TIPS, 5.3% gold, 2.6% commodities, 1.7% equities, 0% REITs. 8 Allocation:
54.1% TIPS, 21.8% commodities, 14.5% REITs, 6.4% equities, 3.3% gold.
Source: FTSE NAREIT All Equity REITs Index, S&P Goldman Sachs Commodity Index, Ibbotson Associates US TIPS, Barclays Capital US Treasury
TIPS Index, S&P 500 Index, S&P GSCI Gold Index.
59
i
i
i
perrucci 2012/7/11 17:12 page 60 #82
INFLATION-SENSITIVE ASSETS
100
90
80
US TIPS
S&P 500
70
60
50
Commodities
40
30
20
Gold
Equity REITs
10
0
5.3
6.1 6.8 7.6 8.4 9.1 9.9 10.6 11.4 12.1 12.9
Portfolio expected inflation-adjusted return (%)
i
i
i
perrucci 2012/7/11 17:12 page 61 #83
Investors using the maximum Sharpe ratio asset strategic allocation, however, would still have been exposed to directional risk
related to the inflation rate (Table 4.5). During low-inflation periods, this strategic asset allocation would have generated real returns
averaging 9.5% per year, with volatility of 12.4% and a Sharpe
ratio of 0.60. During high-inflation periods, however, the maximum
Sharpe ratio allocation would have produced substantially lower
annualised average real returns (5.6%), with only moderately lower
volatility (10.5%) and a Sharpe ratio of 0.48.
Investors seeking to eliminate directional risk altogether (at least
on an expected basis) could have instead chosen a portfolio comprising a 54.1% allocation to TIPS along with 21.8% in commodities,
14.5% in equity REITs, 6.4% in stocks and 3.3% in gold. Across the
entire historical period, this portfolio would have generated real
returns averaging 7.1% per year with 11.3% volatility, for a strong
Sharpe ratio of 0.52; the portfolio would also have provided very
dependable protection against inflation, with nominal returns covering the inflation rate in 75.4% of high-inflation periods. Moreover, the
risk-adjusted returns of this portfolio would not have depended on
the inflation rate: during high-inflation periods real returns would
have averaged 6.0% with 10.4% volatility (Table 4.5), while during
low-inflation periods both real returns (8.3%) and volatility (12.0%)
would have been commensurately higher, resulting in no difference
in returns on a risk-adjusted basis (ie, Sharpe ratios).
i
i
i
perrucci 2012/7/11 17:12 page 62 #84
INFLATION-SENSITIVE ASSETS
81.0
77.8
73.0
71.4
71.4
69.8
69.8
68.3
68.3
61.9
81.0
63.5
58.7
perhaps with monthly lease extensions after the first year, implying that rents can be adjusted annually, if not monthly; self-storage
facilities typically have similarly short lease terms.
For hotel, apartment and self-storage properties, the fact that lease
rents are typically fixed during the entire duration of the contract
term is mitigated by the fact that contract terms are typically short,
potentially enabling property owners to adjust rents in response to
inflation. Other property types may employ automatic adjustments
to changes in inflation, whether explicit or implicit. For example,
many retail leases specify monthly rental payments as a function of
the sale revenues generated by each store; thus, as inflation affects
the sale prices of consumer goods, it affects lease rents as well.
In some cases, especially with tenants that are government agencies, office leases may include an explicit adjustment in response to
inflation; in these cases, office property returns may be sensitive to
inflation even with long lease terms. A slightly different mechanism
may affect lease rents for health-care properties, including long-term
care facilities: if health-care reimbursement rates are regulated, then
inflation may be accounted for in determining payments for various health-care services, and therefore pass through to owners of
health-care properties.
Table 4.6 presents the same analysis of the inflation-protection
effectiveness shown in Table 4.1, but focuses on publicly traded
62
i
i
i
perrucci 2012/7/11 17:12 page 63 #85
equity REITs across different sectors over the historical period from
January 1994 to August 2011 for which data is available.15 As
Table 4.6 shows, the two property sectors that provided the most
effective inflation protection, with returns greater than or equal to
inflation during high-inflation semesters, were self-storage (81.0%)
and residential (77.8%), that is, two of the property types characterised by short lease terms, and therefore frequent lease turnover
and renegotiation. Lodging, however (the third property type characterised by short leases), demonstrated a success rate of just 68.3%,
less than the equity REIT industry as a whole (71.4%). Shopping
centres, too, provided inflation-protection dependability above that
of the industry as a whole, at 73.0%, but the other two retail property types, regional malls (69.8%) and especially free-standing retail
(61.9%), fell short of the industry average, as did other property
types including office (69.8%) and health care (68.3%).16
INFLATION HEDGING WITH ILLIQUID REAL ESTATE
INVESTMENTS
We have measured real estate investment returns using the returns
on publicly traded equity REITs, but several other real estate
indexes are available, including the NCREIF Property Index (NPI) of
unleveraged core property returns published by the National Council of Real Estate Investment Fiduciaries, the Open-End Diversified
Core Equity (ODCE) Fund Index, also published by NCREIF, and
indexes of private equity real estate fund investments published
jointly by NCREIF and The Townsend Group.17 All of these indexes
measure the returns of illiquid investments, either in commercial
properties themselves (held directly or through separate accounts
with investment managers) or in non-traded shares of private equity
real estate funds.
The illiquidity of such real estate investments means that their
periodic returns, unlike the returns of publicly traded REIT equities,
cannot be measured directly on the basis of price discovery from
actual transactions.18 Instead, returns on unlisted real estate investments are appraisal based, ie, appraisals are conducted (whether
internally or externally) and used to periodically (typically quarterly) estimate the capital appreciation component of total returns.
For the purpose of evaluating non-traded real estate holdings as an
63
i
i
i
perrucci 2012/7/11 17:12 page 64 #86
INFLATION-SENSITIVE ASSETS
Return component
Measure
Correlation with
quarterly CPI (%)
Capital appreciation
NPI
ODCE
31
31
Income return
NPI
ODCE
17
27
Total return
NPI
ODCE
TBI
32
33
10
inflation hedge, this introduces a critical problem, as current or estimated inflation is a typical input in the appraisal process, or in the
extrapolation between appraisals. Thus, comparing appraisal-based
returns with inflation as a means of analysing sensitivity to inflation
becomes tautological.19
Circumstantial evidence for this problem can be seen in Table 4.7,
which shows the correlation between quarterly inflation and quarterly capital appreciation and income returns on illiquid real estate,
as measured by the NPI and the ODCE. For both indexes, correlation between capital appreciation and inflation (31%) is greater than
correlation between income and inflation (27% from the ODCE, and
just 17% from the NPI). This suggests that the appraisal-based values of commercial properties fluctuate more strongly in response
to inflation than does the quarterly income produced by the same
properties (as measured by actual rents received).
Additional evidence is uncovered by comparing correlations
computed from appraisal-based total returns with correlations computed from actual transaction values. Table 4.5 also shows the correlations of quarterly inflation with quarterly total returns measured
by the NPI, the ODCE and the Transaction Based Index (TBI) calculated until recently by the Center for Real Estate at the Massachusetts Institute of Technology. The TBI is computed using all the
transactions of properties that are also in the database used to compute the NPI, which itself has substantial overlap with the database
used to compute the ODCE, implying that differences among the
64
i
i
i
perrucci 2012/7/11 17:12 page 65 #87
assets underlying the three indexes are slight. The computed correlations, however, are different: just 9.8% using actual transaction
values as measured by the TBI, compared with 32% using appraised
values from the NPI, and 33% using appraised values from the
ODCE.
To summarise, we caution against the use of appraisal-based estimates to evaluate the sensitivity of asset returns to inflation. While
the evidence reviewed here is specific to the real estate asset class
in the US, the conclusion seems equally applicable to other illiquid assets whose values and returns are estimated by appraisal. If
valuations are influenced in part by the appraisers awareness of
current or expected inflation, then evaluating inflation sensitivity
may amount to tautology.
CONCLUSIONS
According to the Gordon growth model, real estate can be considered
a perfect hedge against inflation, under the strong assumption that
future rent growth and discount rates move in line with expected
and actual inflation rates. In this chapter, we have examined the
historical performance of real estate as an inflation hedge from 1978
to 2011, and compared it with other inflation-sensitive asset classes.
In the historical sample, looking at single asset classes first, commodities provide the best inflation protection, as per the measure
of hedge effectiveness adopted here, with an overall success rate of
70% in high-inflation semesters (75% for energy commodities and
61% for non-energy commodities). These results are only slightly
sensitive to differences in the time horizon used to calculate returns,
the demarcation line used to define high-inflation periods and the
choice of synthetic TIPS return series.
During low-inflation periods, however, commodities generated
the lowest returns of any asset class considered. This large performance difference highlights the utility of constructing a balanced
portfolio if performance in both high- and low-inflation regimes is
the goal.
Historically, a Markowitz meanvariance optimisation suggests
that a blended portfolio, invested 49% in TIPS, 17% in equity REITs,
15% in commodities, 14% in stocks and 6% in gold, achieves the
65
i
i
i
perrucci 2012/7/11 17:12 page 66 #88
INFLATION-SENSITIVE ASSETS
maximum Sharpe ratio (0.54) across all semesters in our sample. The
success rate of this multi-asset-class portfolio is quite high (75%), but
it also has considerable directional risk.
To mitigate the latter, historically, slightly more could have been
invested in TIPS (54%) and commodities (22%), and slightly less
in equity REITs (14%), stocks (6%), and gold (3%). Historically, this
portfolio has provided not only a similar success rate (75%) in highinflation semesters, but also a similar Sharpe ratio (0.52), with the
advantage that the latter is identical in both high- and low-inflation
periods.
Finally, investors seeking to maximise the success rate in highinflation semesters, without regard to directional risk, would have
chosen a more aggressive portfolio, with 55% in commodities, 39%
in REITs, 5% in TIPS, 1% in stocks and no gold holdings. Historically,
this portfolio has a success rate of 78% in high-inflation semesters,
but considerable directional risk.
Different property types provide different levels of inflation protection, depending on the extent to which rents adjust to inflation. The property types expected to provide the strongest inflation
protection are those characterised by short-duration leases, or by
rents linked to revenues. Empirical data generally supports these
expectations, with self-storage, residential properties and shopping
centres having a success rate ranging from around 75% to 80%
in high-inflation semesters (Table 4.6), higher than the industry
average (71%).
Although we have used publicly traded equity REIT returns, similar empirical analysis could in principle be conducted using returns
on illiquid investments, ie, properties themselves or private equity
real estate investment funds. Unfortunately, the latter are typically
estimated by appraisals, which are linked to inflation, thus making
an analysis of their price sensitivity to inflation amount to tautology.
The empirical evidence examined in this chapter suggests that a
variety of assets have inflation-protecting characteristics. Real estate,
considered a strong inflation hedge on conceptual grounds, has in
fact performed as well as, or better than, other inflation-sensitive
assets in the historical sample considered, and has not exposed
investors to significant directional inflation risk. Indeed, based
on both empirical results and theoretical arguments, real estate,
accessed through publicly traded equity REITs, provides attractive
66
i
i
i
perrucci 2012/7/11 17:12 page 67 #89
Strategic asset allocation is the choice of a set of portfolio weights that are not expected to
change in response to market conditions, requiring only that the portfolio be rebalanced periodically to the strategic weights. Tactical asset allocation is the choice of portfolio weights in
response to current market conditions, overweighting assets or asset classes that are expected
to outperform, while underweighting those expected to underperform.
Discussing the limitations of this simple valuation model is outside the scope of this chapter.
For a discussion with reference specifically to real estate and REITs, see Geltner et al (2007).
Clearly, this is a very strong assumption that may not generally hold in reality, but it is a
useful baseline for illustrating the relevant issues.
Demand shocks also affect real interest rates in the short run.
See http://www.snl.com.
If the analysis is restricted to the historical period from January 1990 to September 2008,
thereby excluding the liquidity crisis of 20089, then during months of higher-than-median
inflation the median return of high-leverage REITs (1.14%) is slightly less than the median
return of low-leverage REITs (1.17%); during months of lower-than-median inflation, returns
to high-leverage REITs were more markedly less than returns of low-leverage REITs (1.34%
versus 1.92%).
That is, month n principal is linked to an interpolated value of the CPI published in months
n 1 and n 2, which measure inflation in months n 2 and n 3, respectively. Moreover,
TIPS income is paid only every six months, a consideration relevant to some retail investors.
For example, we might define the most effective hedge as the one that minimises the variance
of the overall position. In this case, if series Y and X have correlation and volatilities Y
and X , respectively, the optimal hedge ratio is given by h = X /Y , which means that to
obtain the optimal (minimum variance) hedge we need to sell h units of asset Y for each unit
of X.
10 January 1978 is the first date for which the S&P GSCI Gold Index is available. Data for the
other assets is available from January 1972 (the starting date of the FTSE NAREIT Equity
REITs Index); over the longer historical period the success rate was about 67% for both REITs
and commodities, and about 55% for both stocks and TIPS.
11 Although TIPS income return is linked to, and thus increases with, inflation, the TIPS Index
also has real rate duration, ie, its price decreases with an increase in real rates. This effect
is especially important before TIPS first issuance, where real rates have no observable market dynamics, and are reconstructed by subtracting realised inflation to market-observable
nominal rates. Since nominal rates typically increase more than one-to-one with inflation,
these historically backfilled synthetic real rates will also tend to increase with inflation and
decrease with index price, thus underestimating the hedging performance of the asset class
in high-inflation scenarios.
12 This ranking may seem intuitive given that higher-volatility, higher-return assets such as
commodities, REITs and equities will have a greater chance to satisfy our inflation-protection
criterion during high-inflation periods. As noted in the next section, however, this intuition
does not hold firmly. The reader should bear in mind that the measured performance of TIPS
may be sensitive to the methodology used to construct a synthetic TIPS return index for
periods preceding TIPS issuance (ie, before 1997), as discussed in the next section.
13 This is mirrored by the fact that commodities become the worst performing asset in lowinflation months.
67
i
i
i
perrucci 2012/7/11 17:12 page 68 #90
INFLATION-SENSITIVE ASSETS
14 The Sharpe ratio is defined as the ratio of portfolio return in excess of the risk-free rate to
portfolio volatility. In the analysis, the monthly risk-free rate is given by the return on the
Citigroup BIG 1-month US Treasury Bill Index.
15 The historical period for the analysis of inflation-protection dependability by property type
begins in January 1994, the inception date for the FTSE NAREIT family of property-type
indexes.
16 Other property types for which separate historical data is not available include timberland
and specialised data centres.
17 See http://www.ncreif.org.
18 Transaction-based indexes of commercial property values and returns do exist, including the
TBI employed later in this section. Several researchers, including Lin and Vandell (2007),
Cheng et al (2010) and Bond and Slezak (2010), however, have noted that transaction-based
index methodologies applied to illiquid assets provide biased measurements of both the
average and the volatility of returns or changes in value.
19 The same, or an analogous, problem may affect the backfilling of returns on TIPS prior to
December 1997.
REFERENCES
Bhardwaj, G., D. J. Hamilton and J. Ameriks, 2011, Hedging Inflation: The Role of
Expectations, Report ICRUIHE 042011, Vanguard Group, URL: https://www.vanguard
investments.de/content/documents/Articles/Insights/hedging-inflation.pdf.
Bond S. A., and S. L. Slezak, 2010, The Optimal Portfolio Weight for Real Estate
with Liquidity Risk and Uncertainty Aversion, Working Paper, URL: http://ssrn.com/
abstract=1691503.
Cheng, P., Z. Lin and Y. Liu, 2010, Illiquidity and Portfolio Risk of Thinly Traded Assets,
Journal of Portfolio Management 36(2), pp. 12638.
Geltner, D. M., N. G. Miller, J. Clayton and P. Eichholtz, 2007, Commercial Real Estate:
Analysis and Investments, Second Edition (Mason, OH: Thomson South-Western).
Gordon, M. J., 1962, The Investment, Financing and Valuation of the Corporation (Homewood,
IL: Irwin).
Lin, Z., and K. Vandell, 2007, Illiquidity and Pricing Biases in the Real Estate Market,
Real Estate Economics 35(3), pp. 291330.
Lomelino, D., K. Gillett and M. Komarynsky, 2011, Inflation Hedging with InflationLinked Bonds, Report, Towers Watson, URL: http://www.towerswatson.com/assets/
pdf/4125/1101-TIPS-FIN.pdf.
Markowitz, H., 1952, Portfolio Selection, The Journal of Finance 7(1), pp. 7791.
Markowitz, H., 1959, Portfolio Selection: Efficient Diversification of Investments (New York:
John Wiley and Sons).
Ralls, B., 2010, Inflation or Deflation: Prepare for Either, Fidelity Investments, URL:
https://guidance.fidelity.com/viewpoints/inflation-vs-deflation.
68
i
i
i
perrucci 2012/7/11 17:12 page 69 #91
i
i
i
perrucci 2012/7/11 17:12 page 70 #92
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 71 #93
As mentioned above, there are three separate segments that fit our
definition of infrastructure assets. We will discuss each of these, and
their unique features, in the following.
ENERGY AND WATER UTILITIES
Utilities that meet our previously stated definition of infrastructure
include the following:
utilities that engage in transmission and distribution4 of elec-
tricity; these companies are responsible for either the transport of electricity from power generators to communities on
high-voltage power lines or for electricity distribution within
communities on low-voltage power lines;
utilities that engage in the transmission and distribution of gas;
Because of the vital public and safety service provided, these companies have a natural monopoly and are generally heavily regulated,
with the government monitoring and controlling the price charged
for their services. In essence, the relevant government regulator
grants the privilege (and responsibility) to deliver reliable, quality
service and, in exchange, it sets a pricing mechanism to secure a fair
rate of return for the utility company.
The price-setting process requires the regulator to take into
account the impact of several effects, including higher interest rates
(which, as mentioned before, affect the cost of debt and equity capital) and higher inflation (which increases operating costs and construction costs on new assets built, as well as the value of the assets
already owned). As a consequence, the value of an efficiently regulated utility company, as measured by the net present value of
its cashflows, should be relatively stable, irrespective of economic
conditions, and inflation in particular.5
In practice, the protection to earnings from inflation depends upon
the frequency of the price-setting process, which usually includes a
period of public consultation, and therefore an often sizeable regulatory time lag. Although the latter varies according to different
jurisdictions, price reviews are generally conducted annually, at the
71
i
i
i
perrucci 2012/7/11 17:12 page 72 #94
INFLATION-SENSITIVE ASSETS
Date of
increase
July 2001
April 2003
October 2004
March 2008
September 2010
Toll
increase (%)
17.0
14.0
13.0
8.6
8.6
Year
+9.2
+5.8
+5.5
+2.5
+2.3
+15.6
+5.0
+4.8
+0.4
+2.5
i
i
i
perrucci 2012/7/11 17:12 page 73 #95
Asset
Location
Basis of
toll increases
407 ETR
Canada
APRR
Atlantia
Brisa
Chicago Skyway
France
Italy
Portugal
US
CityLink
Australia
Greater of
4.5% or CPI
to 2015 then
CPI
Quarterly
Eastern Distributor
Australia
Greater of
4.1% or
basket of 67%
average
weekly
earnings and
33% CPI
In A$0.50
increments
US
Greater of
2%, CPI or
nominal GDP
per capita
CPI
At operators
discretion
Annually
M5
M6 Toll
Western Harbor
Tunnel
Australia
UK
Hong Kong
At operators
discretion
85% of CPI
70% of CPI
90% of CPI
Greater of
2%, CPI or
nominal GDP
per capita
Frequency
CPI
Discretionary
Annually
Annually
Annually
Annually
Annually
Discretionary
Annually
i
i
i
perrucci 2012/7/11 17:12 page 74 #96
INFLATION-SENSITIVE ASSETS
As mentioned above, the basis on which tolls are increased is controlled by the terms of the concession agreement. There are only
three toll roads of any significance in the Western world where the
concessionaire has full discretion on toll increases: the M6 Toll in
the UK, the 407 ETR in Canada and the South Bay Expressway in
California. All other tollroad operators have to follow specific formulas, which are generally linked to inflation. Table 5.2 provides
an illustrative cross-section. As can be seen, the pricing mechanism
for these tollroads picks up any increases in inflation, with minimal
lag. Consequently, the majority of tollroad operators have the ability to respond quickly to any spike in inflation. And as the data in
Table 5.1 highlights, tollroad concessionaires can expect that there
will be minimal, if any, disruption in demand as tolls are increased,
so revenues, which are the product of toll price and traffic volume,
will fully offset the inflationary impact.
Of course, the other key area where inflation can have an impact
is on capital expenditures. With most tollroads, however, capital
expenditures are minimal, and generally limited to keeping the
roads in good operating conditions, by resurfacing, replacing ageing
crash barriers, etc, as needed. Consequently, for this infrastructure
segment, inflation has very little material effect when it comes to
capital expenditures. Another crucial aspect is that tollroads, like
most infrastructure assets, are generally more highly leveraged than
average industrial companies. The impact of higher inflation on debt
costs is important, and it is covered later in the chapter.
AIRPORTS
Airports involve two separate businesses, ie, airside and landside
operations.
Airside operations encompass the management of the aeronautical aspects, such as the operation and maintenance of the runways
and taxiways of the airport. The majority of airside revenue is generated by a charge levied on passenger movements (ie, passenger
arrivals and/or departures), a charge levied on each aircraft movement or a combination of both. In most jurisdictions, the onus is on
the airport to negotiate appropriate charges with the airlines, with
some form of regulation as a fall-back position. This component
of the airports operations therefore behaves much like a regulated
utility, with increases in inflation leading to increases in charges in
74
i
i
i
perrucci 2012/7/11 17:12 page 75 #97
order to ensure the airport earns a fair rate of return on the capital
invested.
Landside operations involve the remainder of the airport, and fall
into three primary areas: retail operations, car parking and general
property development. Most airports do not directly run the retail
outlets. Instead, they act as the lessor and receive a guaranteed minimal rental, normally indexed to inflation, along with a share of sales.
Hence, a spike in inflation will lead to an increase in nominal retail
sales (assuming static demand), and the airport will benefit from the
inflation protection of these revenues.
The car parking operations at the airport generally behave like a
monopoly, although there is some substitution threat, ie, the potential for travellers to park off airport, or to use alternatives, such as
taxis or public transportation. However, the level of substitution
threat is generally weak, and thus the airport has significant pricing
power that can be used to offset inflationary spikes.
As for property development, there is a range of businesses that
seek to be located on or close to airports, including the airlines
and freight forwarders, along with the customs and immigration
officials. Accordingly, airport operators typically invest in property
assets and act as the lessor, receiving a rental stream that is indexed
to inflation, thus offering protection against rises in price levels.
Airports incur the highest level of capital expenditures among
the infrastructure segments. Airside capital expenditures include
widening and extension of runways and taxiways, and are generally undertaken after consultation and agreement with the airlines
as well as the relevant regulatory authorities. Airside charges are
typically increased to recover these development and maintenance
costs over time. Landside capital expenditures are often incurred
to increase the retail or parking space available, or other property
leasing facilities. Higher inflation may affect the financial viability of
such capital expenditure, but it is expected that the airport operator
will be rational and elect not to proceed with such expansion projects
if the project is not expected to yield a reasonable real rate of return.
In other words, if project and revenue planning is done correctly,
inflation will increase the cost of capital expenditures and yet have
a minimal impact on the value of the airport asset as a whole.
When considering an airport asset as an investment, there are several issues that need to be carefully considered, as they contribute
75
i
i
i
perrucci 2012/7/11 17:12 page 76 #98
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 77 #99
simplify their cash-management position, as they will be able to service their debt with stable fixed-rate outflows, even if interest rates
rise. In addition, the net-present value of their long-term fixed-rate
liabilities will decrease if interest rates and inflation increase, thus
offsetting, at least in part, the imperfect inflation-linkages on the
revenue side.
CONCLUSIONS
Although infrastructure is a relatively new asset class for many
investors, it is increasingly seen as an attractive substitution alternative in the context of a diversified portfolio. When it comes to
inflation protection in particular, several characteristics provide an
attractive platform for investors seeking a safe haven, including
the fact that infrastructure revenues are typically structurally linked
to inflation and relatively high levels of debt are generally managed
prudently in order to minimise, or even benefit from, interest rate
movements.
1
Although commodity price risk is definitely a cause of earnings volatility, the link between
commodity prices and inflation is weak at best. This is why, in the example that follows, we
do not consider merchant power generators to be infrastructure assets, as their earnings are
business cyclical, and yet not structurally linked to inflation.
Here we assume the debt is floating rate and resets periodically. Note that, in this case, if
inflation and interest rates rise, the cashflow outlays required to service the debt will increase,
but the debts net present value will remain unchanged.
This depends on the price-transferring mechanisms, the structure of operating costs and the
nature of liabilities.
Transmission refers to the transport of the commodity from source to the distribution hub, eg,
from the electricity generation plant to an electricity sub-station in a community. Distribution
is the transport of the commodity from the distribution hub to the point of use, ie, homes,
offices, factories, etc.
In other words, for the value of an infrastructure asset to be stable, earnings should grow with
inflation, so that their net present value remains unchanged.
Toll increases usually occur in sizeable increments and in a highly publicised manner for the
Eastern Distributor (Table 5.1), as tolls are paid in cash. Fully electronic tollroads, like the
CityLink in Melbourne (Australia), the Westlink M7 in Sydney (Australia) and the 407 ETR
in Toronto (Canada) have tolls regularly increased by lower increments, with little fanfare,
and no impact on demand.
This is true for several airports in Europe, while the issue is typically not present in Australia
and New Zealand.
77
i
i
i
perrucci 2012/7/11 17:12 page 78 #100
i
i
i
perrucci 2012/7/11 17:12 page 79 #101
i
i
i
perrucci 2012/7/11 17:12 page 80 #102
INFLATION-SENSITIVE ASSETS
Table 6.1 Carmike Cinemas, Inc: price changes versus consumer price
inflation (19922010)
Year
Admission
(US$)
Concessions
(US$)
Total
(US$)
1992
2010
3.47
6.85
1.28
3.43
4.75
10.28
3.9
5.6
4.4
2.5
CPI, All Urban Consumers, US City Average (all items); not seasonally
adjusted.
Source: data from Department of Labor, Bureau of Labor Statistics.
We shall illustrate this point by using Carmike Inc as the business model typical of cinema chains (Table 6.1).1 Historically, cinemas have had sufficient pricing power to raise ticket and concession
prices at rates exceeding general inflation price indexes. This pricing power has persisted even though this particular business sector
experienced some degree of saturation during the period considered, when the market might have been expected to revert to a pricecompetitive system (cinema attendance in North America peaked in
2002).
Even modest pricing power (Table 6.2) is important, as it can
be leveraged against a relatively fixed cost structure, resulting in
a meaningfully greater expansion in net income (which, simply put,
means that if most of your costs are fixed, but you can increase your
prices, then net income (revenue minus costs) will increase). A cinema chain is a good example of a high-fixed-cost, low-marginal-cost
business. Some 85% of the operating expenses of a cinema are relatively fixed: basic operating expenses such as rent and film exhibition
costs. Costs that might be considered more variable with respect to
patronage volumes, administrative expense and the cost of operating the food concessions amount to only about 3% each. Revenues
from the food concessions, though, comprise about one-third of revenues, the balance being ticket sales. Accordingly, the change in operating costs between a cinema operating at 50% of capacity and 95%
of capacity is virtually nil (the same rent must be paid, the same
80
i
i
i
perrucci 2012/7/11 17:12 page 81 #103
2001
Avg ticket price (US$)
Avg concession sales
per patron (US$)
Total attendance (mn)
4.83
2.10
64.261
2004
Total Annualised
change change
(%)
(%)
5.17
2.33
7.0
10.9
2.3
3.5
63.260
(1.6)
(0.5)
494.5
(423.6)
70.9
(26.1)
44.8
9.1
8.2
0.7
95.9
186.8
65.3
minimal staffing is required for the ticket line and concession stands,
and so forth) so that the earnings are highly sensitive to capacity utilisation. The alluring income margin growth rate reported in this table
should not be misinterpreted as representative of the companys (or
the cinema industrys) long-term results, since the 20014 period
was selected specifically to illustrate this form of operating leverage. From this example, it can be seen that, even with slightly lower
attendance at cinemas, a modicum of pricing power, of the order
of 3% per annum, was sufficient to raise pre-tax income by 65.3%
(Table 6.2).
All else being equal, an increase in ticket and concession pricing,
since it requires no increase in operating cost, amounts to a pure
increase in operating profit. If the initial profit margin is relatively
low (in this case about 6%), then pricing power of merely 23% per
year can be sufficient to increase the profit margin by around 50%
or more in a very few periods.
Therefore, equities as a class do have the ability to mitigate
the impact of inflation on purchasing power through the earnings
leverage of a business. Unfortunately, this does not mean that an
investor can automatically capture the inflation effects on earnings
as reported in a companys income statement. Even specific equity
81
i
i
i
perrucci 2012/7/11 17:12 page 82 #104
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 83 #105
rose with inflation, thereby limiting the potential for margin expansion. During the period, Newmonts share price only increased by
an annualised 2.2% in nominal terms, for a negative return in inflation adjusted terms (annualised consumer price inflation was 7.4%
in the decade from 1970 to 1979).
The key variable in this example was the rising earnings yield for
shares driven by rising interest rates, among other factors, as well
as the negative effect on operating margin due to higher replacement costs. Similar counter-intuitive examples can be found in oil
exploration and drilling companies during periods of rising oil
prices.
The impact of expectations on prices
Another important element is that, in addition to inflation and interest rates, market prices embed expectations about future earnings
and other cashflow streams, and these expectations, whether confirmed by future economic realisation or not, powerfully influence
where stocks trade, over longer periods of time than we have been
taught to believe.
To illustrate this point, we shall examine how Newmont stock performed from 1996 to 2010, again using the literary conceit of perfect
foresight. Say that it could have been known in 1996 that gold prices
would more than triple over the ensuing 14 years (from an average price of US$388 per ounce in 1996 to US$1,216 per ounce in May
2010, for an annualised appreciation rate of 8.5%). As a counterpoint
to the preceding example, though, it is also known that consumer
price inflation would average a mere 2.4%, lower than any year during the prior decade, and that interest rates would actually decline.
Nevertheless, Newmont stock, which traded at US$60.25 per share in
May 1996, was lower, at US$53.82 per share, in May 2010. Therefore,
a seemingly obvious investment in Newmont would have resulted
in a loss in nominal terms and, more importantly, a substantial negative real rate of return over the period considered, despite much
higher gold prices and lower interest rates.
In this example, the limiting variable on the share price was not
inflation or higher interest rates and, consequently, a lower P/E ratio
(higher required earnings yield). Rather, in the 1990s, gold mining companies did not produce significant earnings, so investors
tended to value them based on net asset value. This involves estimating production over the life of the mine (several decades forward),
83
i
i
i
perrucci 2012/7/11 17:12 page 84 #106
INFLATION-SENSITIVE ASSETS
projecting the prices at which gold will be sold in the future and subtracting the estimated cost of production to derive the prospective
annual payment streams. A net-present-value calculation is applied
to each of those payment streams, the sum of which would equal,
after allowance for net cash or debt, the net asset value of the company. Using that methodology, at a gold price of US$388, investors
concluded, in aggregate, that fair value in 1996 was US$60 a share.
Yet, by 2010, when mining companies were producing substantial
income, they came to be valued on a price-to-earnings basis, as
opposed to a net-asset-value basis, by which approach investors
concluded that the value of the companys stock was still roughly
US$60, despite gold prices being dramatically higher.
The lesson to be learnt is that even when the earnings of companies
benefit from inflation, if too many investors anticipate a positive
case outcome, that outlook will be largely discounted in their stock
prices, the result being a poor inflation hedge. If the initial valuation
of such a company is too high, there is hardly a realistic level of rapid
earnings growth, even a decade of such growth, that will suffice to
overcome the momentary change in clearing prices that can occur
when investors outlooks change.
In addition, many are the complexities of stock and company valuation. For example, in the period considered, the value of Newmont
as a company, that is, its market capitalisation, did indeed increase
at a double-digit rate. However, this was because of the issuance
of new shares to raise capital. This financing actually diluted the
price per share, which is the only relevant measure to an investor,
contributing to the poor performance outlined above.
The example above highlights how it is not just (inflation) forecasts that count, but their relation to future scenarios embedded in
share prices. Therefore, the true analytical challenge is not merely
finding a company (or type of company) for which a good forecast
can be produced, but finding one that can produce a good forecast
and for which no one else is desirous of, or interested in providing, a good forecast. In other words, the best positioned and best
managed inflation-beneficiary company, if recognised as such and
desired by investors, will paradoxically be priced with a sufficient
premium so as to produce a disappointing return; we must locate an
inflation-beneficiary company whose shares incorporate little or no
expectation of a satisfactory return, because returns are made not on
84
i
i
i
perrucci 2012/7/11 17:12 page 85 #107
i
i
i
perrucci 2012/7/11 17:12 page 86 #108
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 87 #109
Table 6.3 Franco-Nevada and Newmont equity prices versus gold price
Franco-Nevada
Mining Corp
(US$)
Newmont
Mining
(US$)
Gold
(per oz)
(US$)
0.20
23.10
4.23
17.71
382.40
278.95
30.2
8.3
1.7
This is the old Franco-Nevada, for which the earliest share price available is in October 1983, and which was acquired in February 2002 by Newmont Mining. Franco-Nevada became public again in December 2007.
Source: Bloomberg.
i
i
i
perrucci 2012/7/11 17:12 page 88 #110
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 89 #111
Table 6.4 Wilshire US Real Estate Securities Index versus S&P 500
period return (June 2007 to June 2011)
S&P 500
Index
2.6
4.1
increase in revenue requires no direct increase in the number of personnel, computer processing capacity, office space or other expenses
and thus should be particularly profitable.
Real estate companies
Although the real estate industry experienced much distress during
the 20089 financial and housing crisis, this episode has not invalidated real estate as an inflation beneficiary. Indeed, from the beginning of that period up to and including June 2011, real estate stocks
did not underperform broad stock indexes, irrespective of their more
dramatic interim decline. Table 6.4 shows a comparison of returns
between the general US equity market as measured by the S&P 500
Index and the Wilshire US Real Estate Security Index8 , which is a
composite of publicly traded Real Estate Investment Trusts (REITs)
and Real Estate Operating Companies (REOCs).
There is a certain structural limitation to the ability of REITs to
compound their per-share value, since by regulation they must distribute as dividends substantially all of their income. As a result,
they must issue additional shares in order to acquire new properties, and there are periods when the cost of new equity capital
will be disadvantageous. Non-REIT real estate companies can reinvest their income and are less dependent upon the public market
for growth capital. Therefore, we might be better advised to concentrate on select commercial real estate companies, with strong balance
sheets and unique income-producing properties. Aside from their
fundamental investment merits, these companies provide inflationhedging benefits through multiple modalities. First, inflation typically increases the value of the underlying real estate properties,
so that such a companys net asset value tends to rise in inflationary periods. Second, as their commercial leases approach renewal,
higher prices can be charged to reflect the effects of inflation on
prevailing rental rates. Finally, real estate companies typically use
89
i
i
i
perrucci 2012/7/11 17:12 page 90 #112
INFLATION-SENSITIVE ASSETS
leverage to purchase their properties. To the extent that the mortgage payments are fixed, their liabilities will decrease in real terms
because of higher inflation. In this sense, leveraged companies
in general can benefit an investment portfolio during inflationary
periods.
Leveraged companies and deleveraging mechanisms
Leverage is an important element when evaluating an equity investment. In this section, we shall explain how changes in leverage can
affect the value of a company and thus its stock price. The first mechanism is directly related to inflation and the operational leverage (ie,
higher operating margin) that comes with higher output price levels
relative to a fixed cost structure. The next mechanism is linked not
to inflation per se or to a higher operating margin, but to the expansion in earnings that results from deleveraging, ie, paying down the
debt and reducing interest costs. There is also a market-based mechanism, whereby a companys stock market valuation can rise as a
direct function of its paying down debt. All three mechanisms are
important and can be present at the same time; thus, they are covered
in this section.
We have already discussed how inflation and higher output
prices (versus a fixed cost structure) can result in considerably
higher income margins (Table 6.1). In essence, rising sale prices
provide a form of operating leverage, as they do not require additional resources (such as sales personnel, manufacturing capacity
or other production inputs), and thus do not increase variable costs
(as opposed to rising unit sales). This effect can be amplified by
the presence of fixed-rate leverage as well, since inflation favours
the fixed-rate debtor over the creditor. In that relationship, each
years interest payment remains constant, even as sales revenue
and, presumably, earnings rise in concordance with rising price
levels.
The fact that inflation favours the fixed-rate debtor, due to the
fact that the interest is fixed in the face of overall rising prices, is a
theme that applies beyond publicly traded equities. In fact, one of
the most effective inflation hedges during the inflationary decades
of the 1970s and 1980s was the 30-year fixed-rate home mortgage.
Specifically, from 1973 to 1982, a mortgagors fixed-rate payment
cheapened by 8.7% per annum in inflation-adjusted terms (using the
90
i
i
i
perrucci 2012/7/11 17:12 page 91 #113
average Consumer Price Index (CPI) inflation rate over the period).9
In other words, by the end of the 10-year period, a homeowner would
have been making mortgage payments at 45 = 1/(1.087)10 cents on
the dollar in real terms. In addition, over the same period, they would
have received higher wages and a higher value for their house, in
sympathy with other price levels.
Although fixed-rate leverage amplifies the positive effects on
operating margin of higher output prices, it is the deleveraging of
a company that often offers attractive investment opportunities. In
particular, a company with a high level of fixed-rate leverage, but
in the process of entering a deleveraging phase, might offer the best
of both worlds, ie, attractive performance prospects combined with
protection from rising price levels.
Take the example of a company with a stable business, which
employs its earnings to reduce its debt (through plainly using the
companys earnings to pay down its debt, in the same way that
an individual might use their salary to pay down their credit card
debt).
Initially, we shall assume zero revenue growth (Table 6.5) and
a constant operating margin (no operating leverage) to isolate the
effect of debt reduction. As liabilities are paid down each year,
the interest expense associated with debt is also extinguished. This
leaves more after-tax income available the next year, so that a yet
larger amount of debt may be paid down. This process creates
an increase in net income, which compounds over time (for an
annualised rate of 4.9% in the decade considered).
Table 6.6 presents an additional example, this time with a modest rate (3% per annum) of revenue expansion, but still no operating
leverage (the operating margin is constant at 20%), for an annualised
income growth of 12.7% over the period. In practice, few companies
sustain a linear path of any sort for a whole decade, but the previous examples serve to illustrate the degree to which merely the
two factors of persistent interest expense reduction and the operating leverage of price-inflated revenues can serve the cause of value
creation in a leveraged company.
However, a third and probably the most powerful equity value
creation dynamic from the deleveraging process is expressed by
the MillerModigliani invariance theorems (Modigliani and Miller
1958). These state, in part, that the enterprise value of the firm (that
91
i
i
i
Interest
expense
(US$ mn)
Pre-tax
income
(US$ mn)
Income
net
of tax
(US$ mn)
Debt
(US$ mn)
0
1
2
3
4
5
6
7
8
9
10
Annualised change (%)
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
0.0
200
200
200
200
200
200
200
200
200
200
200
0.0
(150)
(148)
(145)
(142)
(140)
(137)
(133)
(130)
(127)
(123)
(120)
(2.2)
50
52
55
58
60
63
67
70
73
77
80
4.9
33
34
36
37
39
41
43
45
48
50
52
4.9
2,000
1,968
1,933
1,898
1,860
1,821
1,780
1,736
1,691
1,643
1,594
(2.2)
Interest
coverage
ratio
1.33
1.36
1.38
1.41
1.43
1.46
1.50
1.54
1.58
1.62
1.67
Year
Revenue
(US$ mn)
Operating
income
(US$ mn)
INFLATION-SENSITIVE ASSETS
92
Table 6.5 The effects of deleveraging with no revenue growth/no operating leverage
We make the following assumptions. Revenue growth: 0%. Operating margin: 20%. Debt interest rate: 7.5%. Income tax rate: 35%.
i
i
i
Table 6.6 The effects of deleveraging with modest revenue growth/no operating leverage
Pre-tax
income
(US$ mn)
Income
net
of tax
(US$ mn)
Debt
(US$ mn)
0
1
2
3
4
5
6
7
8
9
10
Annualised change (%)
1,000
1,030
1,061
1,093
1,126
1,159
1,194
1,230
1,267
1,305
1,344
3.0
200
206
212
219
225
232
239
246
253
261
269
3.0
(150)
(148)
(145)
(141)
(138)
(133)
(129)
(123)
(117)
(111)
(103)
(3.7)
50
58
67
77
87
98
110
123
136
150
166
12.7
33
38
44
50
57
64
72
80
88
98
108
12.7
2,000
1,968
1,930
1,886
1,836
1,779
1,715
1,643
1,563
1,475
1,377
(3.7)
Interest
coverage
ratio
We make the following assumptions. Revenue growth: 3%. Operating margin: 20%. Debt interest rate: 7.5%. Income tax rate: 35%.
1.33
1.40
1.47
1.55
1.64
1.74
1.86
2.00
2.16
2.36
2.60
Interest
expense
(US$ mn)
93
Year
Revenue
(US$ mn)
Operating
income
(US$ mn)
i
i
i
perrucci 2012/7/11 17:12 page 94 #116
INFLATION-SENSITIVE ASSETS
is, the sum of its debt and equity) is invariant (under certain assumptions) to the capital structure of the firm. This presumes that companies are valued in the marketplace on an enterprise value basis,
irrespective of the distinction between debt and equity. Accordingly,
if a company has a low stock market capitalisation in relation to its
debt, and if it uses its cashflow to reduce debt, the equity value
should increase by a like amount such that the total enterprise value
remains unchanged. As a conceptual test, were this not so, then a
highly leveraged company that repays debt would find its enterprise
value contracting, which is to say that investors would penalise the
company for reducing its financial risk.
In our experience, deleveraging companies exhibit such distinctive characteristics that they deserve their own sector classification.
In other words, companies in separate sectors yet with common
balance-sheet structures and debt reduction strategies will share
greater return commonalities than many companies in the same
sector but with different balance-sheet arrangements (however, the
unfolding of events over a span of years is rarely cleanly reducible
to the single variable). A clean, simple example of debt reduction
without any meaningful intrusion of other balance-sheet/operating
issues, would be Church & Dwight. However, in this case, its starting
point was leveraged only on a balance-sheet basis, not on an income
statement/interest coverage basis. Therefore, it was not obviously
undervalued at the start, and its share price appreciation over time
was more a function of its business expansion than its balance-sheet
improvement (Table 6.7).
As shown in Tables 6.5 and 6.6, investing in companies that are
entering a deleveraging phase can provide very rewarding returns.
An additional benefit from investing in deleveraging firms is that
returns tend to be less correlated to exogenous events such as the
economic cycle or the competitive environment, since the equity
value expansion is largely a function of the debt reduction process
itself. Furthermore, these companies have typically established their
competitive positions already, and do not need to incur the costs (and
uncertainties) required to gain market share or develop a new product. Clearly, the objective should be to select companies with stable
businesses, good management and for which leverage has not been
the result of deteriorating fundamentals, but the most effective strategy to finance the business, and enhance returns on equity. As with
94
i
i
i
Sep 2011
Dec 2010
Dec 2009
Dec 2008
Dec 2007
Dec 2006
income.
Book
value
(US$ mn)
Debt/
equity
(%)
Sales
(US$ mn)
254
340
816
856
856
933
2,086
1,871
1,602
1,332
1,080
864
12
18
51
64
79
108
2,704
2,589
2,521
2,422
2,221
1,946
Interest
expense
(US$ mn)
10
28
36
47
59
54
Operating
income
(US$ mn)
Interest
savings
(%)
Interest
coverage
ratio
Book value/
share
(US$)
457
445
413
340
305
252
142
24
16
34
9
45.8
16.0
11.6
7.2
5.2
4.7
29.10
26.34
22.76
19.62
16.41
13.31
operating income and interest expense for 2011 based on nine-month results, annualised.
As
95
Sales,
Total
debt
(US$ mn)
i
i
i
perrucci 2012/7/11 17:12 page 96 #118
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 97 #119
of the stock while retaining the interest income stream, final maturity (unless a perpetual preferred, ie, a preferred stock with no final
maturity) and superior capital position and legal rights of traditional debt. For these reasons, convertibles are conventionally used
as a conservative equity substitute.12
Yet, in practice, when a convertible security is sold, two premiums
are paid: a premium to the common share price13 and the share price
itself, which is likely to have already appreciated to a fair degree.
Moreover, in comparison with other bonds, it provides a belowmarket yield as a trade-off for its optionality. Accordingly, whatever
the future return prospects might be, we cannot say that, on a probabilistic basis, the riskreward trade-off favours the buyer. Moreover,
in comparison with other bonds, it provides a below-market yield as
a trade-off for its optionality. Accordingly, whatever the future return
prospects might be, we cannot say that, on a probabilistic basis, the
riskreward trade-off favours the buyer. If for no other reason than
mean reversion behaviour, much less the normal vagaries of the market, there is a reasonable (perhaps more than reasonable) chance that
the underlying shares may decline sharply at some point.
If the shares do decline sharply, the bond, which was originally
priced relative to the share price, will fall as well. So long as creditworthiness is not at issue, the bond should decline only to a price
that provides the same 7% yield at which the company would have
had to issue a conventional bond. It is now a busted convertible
and it still retains the embedded equity option, although this will
have fallen very far out-of-the-money. It now has more intriguing
characteristics.
Let us observe an example of such a cycle from start to finish. Consider XYZ Corporation (XYZ), a well-regarded growth14 company
with a BBB credit rating, which is considering obtaining financing in
the market for a term of 10 years. Given its credit rating, XYZ could
issue a traditional bond (at par), paying an interest rate of 7% per
annum. Fortunately, the XYZ shares have recently appreciated from
US$20 to US$30. Let us say that US$20 was an appropriate valuation at the time, based on generally accepted metrics, and that US$25
was appropriate enough as a forward value, but that US$30 could
be deemed excessive.
XYZ Corp capitalises on this opportunity by issuing a con-
i
i
i
perrucci 2012/7/11 17:12 page 98 #120
INFLATION-SENSITIVE ASSETS
or by 50%.
The price of the convertible security at year 1 is still the sum of
notes are no longer suitable. They served one purpose in having outperformed the underlying stock in the decline but, at
a 117% conversion premium, they no longer contain much
optionality or equity sensitivity. The shares would have to
nearly double in value for the convertible equity sensitivity
to be restored to levels similar to those at inception.
However, at this same point the busted convertible does have
intriguing characteristics from the perspective of inflation mitigation, depending upon the characteristics of the issuing company and
its equity. Used in this way, busted convertibles function more as
bond substitutes than equity substitutes. In the favourable case, the
assessment might be as follows.
The busted convertible note now has the full bond yield that
i
i
i
perrucci 2012/7/11 17:12 page 99 #121
US$48.91;18
99
i
i
i
perrucci 2012/7/11 17:12 page 100 #122
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 101 #123
Carmike Cinemas, Inc (ticker CKEC), one of the largest cinema operators in the US is used
here, as it has the longest history of public ownership (according to company 10-K filings). In
fact, Carmike Cinemas has not thrived, though for other reasons related to secular challenges
to this particular business model.
Data for Newmont and Franco-Nevada from 1983 and later is sourced from Bloomberg LP,
which is a generally accepted data aggregator source. Data before 1983 was obtained from
the company. Gold prices are sourced from Bloomberg LP.
Being low-profit-margin intermediaries, interim changes between input costs and output
prices can have a large impact upon year-to-year profitability for grain processors like ADM.
To provide figures more representative of normalised earnings, the three-year average P/E is
used here: 10.4 times for the three fiscal years June 1970 to June 1972; 10.0 times for the three
fiscal years June 1980 to June 1982.
The market capitalisation rate is the rate used to discount future cashflows in the net-presentvalue formula for the share price, or also the market expected yield on the equity investment.
Use of the current prevailing commodity price is typical in these financing contracts.
Often, the royalty contract will contain a schedule such that at the initial gold price, say
US$1,000, the royalty rate might be 2%; at a higher gold price, US$1,100, the rate will be
higher, say 2.5%, and so forth. It is designed in part to limit costs to the miner in the early
production phase, and to generate additional returns for the royalty company in a more
favourable environment for the miner.
http://www.wilshire.com/.
10 These are the credit ratings issued by Nationally Recognized Statistical Rating Organizations
(NRSROs) such as Moodys, S&P, Fitch, etc.
11 This is preferred stock with an embedded option of conversion into common shares. In practice, convertible preferred stock can be seen as a long-maturity convertible bond, but with
101
i
i
i
perrucci 2012/7/11 17:12 page 102 #124
INFLATION-SENSITIVE ASSETS
higher credit risk, given its junior position in the capital structure relative to debt, ie, preferred
stock with an embedded option of conversion into common shares.
12 Depending on market conditions (interest rates, credit spreads and stock prices in particular),
a convertible security might, over time, behave more like a bond or more like a stock, and
thus offer distinct properties and attract interest from different sets of investors at different
times.
13 As mentioned before, the conversion price is typically higher than the current share price.
14 A growth stock is from a company that is expected to grow its earnings and/or revenue faster
than its industry sector or the overall market.
15 The convertible bond can be written as a conventional bond (paying a coupon of c = 5% and
priced at a yield of y = 7%) plus 26.67 equity calls.
16 Assuming a flat and unchanged credit curve for the issuer in question, ie, XYZ Corporation.
In reality, a plunge in stock price would typically imply wider credit spreads.
17 The conversion premium is given by (convertible price/conversion value) 1, where the
conversion value is the value that could be realised by converting into shares immediately.
In our example, the convertible price is US$868, and the conversion value is 26.67 shares
US$15 price per share = US$400. Thus, the conversion premium is (868/400) 1 = 117%. The
higher the conversion premium, the lower the equity sensitivity of the convertible.
18 Average equity appreciation over the nine-year period (from year 1, when the busted
convertible is bought to maturity at year 10) is 14% per annum.
19 This means that, contrary to a convertible bond, there is no maturity date when notional will
be returned.
REFERENCES
Modigliani, F., and M. H. Miller, 1958, The Cost of Capital, Corporation Finance and the
Theory of Investment, American Economic Review 48(3), pp. 26197.
102
i
i
i
perrucci 2012/7/11 17:12 page 103 #125
Inflation-Linked Markets
Gang Hu; Stefania Perrucci
Credit Suisse; New Sky Capital
i
i
i
perrucci 2012/7/11 17:12 page 104 #126
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 105 #127
INFLATION-LINKED MARKETS
i
i
i
Market
No. of
bonds
US
UK
France
31
16
12
Brazil
Italy
13
9
Index
US CPI NSA
RPI
French CPI ex-tobacco
Euro HICP ex-tobacco
IPCA Index
Euro HICP ex-tobacco
Lag
(months)
Coupon
frequency
Deflation
floor
Market
size
(US$ bn)
Maturity
range
(yr)
23
8 or 23
23
Semi-annual
Semi-annual
Annual
Par
None
Par
643
392
222
130
145
130
Up to four weeks
23
Semi-annual
Semi-annual
None
Par
210
143
140
130
CPI, Consumer Price Index; NSA, non-seasonally adjusted; RPI, Retail Price Index; HICP, Harmonized Index of Consumer Prices;
IPCA, ndice Nacional de Preos ao Consumidor Amplo.
Source: data from Bloomberg, Credit Suisse, New Sky Capital. Data is as of June 2011.
INFLATION-SENSITIVE ASSETS
106
Table 7.1 Major government inflation-linked market characteristics
i
i
i
perrucci 2012/7/11 17:12 page 107 #129
INFLATION-LINKED MARKETS
a linker is a real rate product whose market price does not depend
on inflation,5 although its nominal return of course does.
We next discuss the trading and settlement mechanisms for US
TIPS. As a concrete example, consider the US TIPS issued in April
2011, with a 0.125% fixed coupon (semi-annual compounding) and
maturity of April 15, 2016. On Wednesday, October 12, 2011, its
quoted (clean) price was Q = 103 4+.6 Settlement convention is
T + 1, meaning that the trade will settle on Thursday, October 13
(this will be set to t = 0 in the formulas below), when the full (dirty)
price P will be exchanged for the bond. For a US Treasury Inflation
Protected Security, the full price P is given by
P(0) = (Q(0) + AI(0)) IR(0),
where IR(0) =
CPI0
CPIbase
= 0. 060887
2
155
d1
CPId = CPI3mo-lag +
(CPI2mo-lag CPI3mo-lag )
days in month
AI =
(7.1)
(7.2)
i
i
i
perrucci 2012/7/11 17:12 page 108 #130
INFLATION-SENSITIVE ASSETS
Jan
Feb
Mar
Apr
2011
220.223
221.309
223.467
224.906
Year
May
Jun
Jul
Aug
2011
225.964
225.722
225.922
226.545
Date
Days
in
month
3-month 2-month
lag
lag
d 1
14
12
Daily
Reference
Index
30
220.72980
31
226.16316
Index ratio
1.02462
Daily Reference Indexes and Index Ratio are rounded to five decimal
places.
Source: data from BLS, New Sky Capital, Credit Suisse.
T
C IR(0)(1 + BE)k 100 IR(0)(1 + BE)T
+
+ Floor(0)
(1 + n)k
(1 + n)T
k =1
Note that we have written the index ratio in the future (ie, the inflation accretion on the linker notional) as the product of a known
inflation factor at t = 0, ie, the index ratio IR(0), and the future market expectation of inflation, ie, the break-even, over the remaining
life of the bond, eg
IR(k) = IR (0)(1 + BE)k
The deflation floor is struck at par, and its payout at maturity given
by
Floor(T ) = 100 max[1 IR(0)(1 + BE)T , 0]
108
i
i
i
perrucci 2012/7/11 17:12 page 109 #131
INFLATION-LINKED MARKETS
Since the market expectation for inflation, ie, the inflation break-even
(BE), simplifies on the numerator and denominator of the present
value equation, dividing by the known index ratio IR (0), we are
left with a formula that links the quoted price of the linker (Q) plus
accrued interest (AI) directly to real rates
Q + AI =
=
T
C(1 + BE)k 100(1 + BE)T
+
+ Floor (t)
k
T
(
1
+
n
)
(
1
+
n
)
k =1
T
(1 + r)k
k =1
100
(1 + r )T
+ Floor (t)
(7.3)
Neglecting the value of the floor for simplicity,8 the break-even can
be interpreted as the inflation rate that equalises the nominal yield
of the inflation-linked bond with the yield on the nominal bond.
In reality, the bond break-even is not just a pure measure of market
inflation expectations (which are more directly priced in the inflation
swap market covered later in the chapter), as it also contains a liquidity component, which can be material in newly established inflation
markets or during liquidity dislocations such as in autumn 2008.
In any case, this measure has become an important relative value
metric between the nominal and inflation-linked bond markets.
Going back to the linker cashflow formula, note that, since the
linkers coupon is generally smaller than the coupon on the nominal bond of comparable maturity and there is typically a positive
inflation accretion at maturity, the cashflows of the inflation bond
tend to be back-loaded relative to the nominal bond and, as a corollary, the linker exhibits higher rate duration than the nominal bond
of equal maturity (and thus higher credit risk as well). Regarding
duration, Equation 7.3 shows that inflation-linked bonds are sensitive to changes in real rates, while nominal bonds react to changes
in nominal rates (with real and nominal rates typically positively
correlated). Specifically, the duration of a linker, ie, the percentage
change in the full price of the linker for a 1% move in real rates, can
be written as (neglecting the floor)
D=
1 P
1
(Q + AI)
=
P r
Q + AI
r
109
i
i
i
perrucci 2012/7/11 17:12 page 110 #132
INFLATION-SENSITIVE ASSETS
Thus, the linkers duration does not depend on the index ratio, as
the latter simplifies on both numerator and denominator. However,
the dollar change in the linkers price for a basis point move in real
rates (also called DV01) does depend on inflation accretion
DV01 = PD =
(Q + AI)
P
= IR
r
r
This formula shows how the DV01 of a linker is the balance of two
effects that occur over time: a decrease as we move closer to maturity (similar to what happens for nominal bonds) and possibly an
increase in the index ratio (ie, positive inflation accretion over time).
We can also define (real) rate convexity for linkers in a way that
mirrors nominal convexity for conventional bonds. Here, as before,
the index ratio simplifies in both numerator and denominator
C=
1 2P
1
2 (Q + AI)
=
P r2
Q + AI
r2
Finally, a few points on the carry (ie, coupon income minus financing costs) provided by linkers are worthy of mention. Initially, carry
is a function of the fixed coupon (typically lower than the coupon
on the nominal bond of similar maturity) and inflation accretion (ie,
the index ratio), which may partially offset the lower (relative to
nominal bonds) fixed coupon. Returning to the linker cashflow formula, since the latter is indexed to the non-seasonally adjusted CPI,
carry on linkers will not only move with the price index but also
mirror its seasonality fluctuations. Therefore, a (short-term) forecast
of inflation, including seasonal patterns,9 is necessary in order to
calculate carry and is often an important driver of relative value and
break-evens, especially for short maturity linkers.
In the following sections, we discuss how macroeconomic variables, such as real GDP, influence real rates and thus the price of
inflation-linked bonds, and we introduce the different approaches
used by market participants to analyse value in the sector. These
include both non-leveraged and leveraged investors. Typically,
while non-leveraged investors have a long-term horizon and are
mainly driven by the absolute level of real rates, leveraged investors,
who often act as liquidity providers, employ a complementary set
of value metrics and can have an important influence on the market
pricing in the short term. Realistically, most asset managers do end
up playing both roles when managing a linker portfolio, as they will
110
i
i
i
perrucci 2012/7/11 17:12 page 111 #133
INFLATION-LINKED MARKETS
1.0
6 Jan 2010
6 Feb 2010
6 Mar 2010
6 Apr 2010
6 May 2010
6 Jun 2010
6 Jul 2010
6 Aug 2010
6 Sep 2010
6 Oct 2010
6 Nov 2010
6 Dec 2010
6 Jan 2011
6 Feb 2011
6 Mar 2011
6 Apr 2011
6 May 2011
6 Jun 2011
6 Jul 2011
6 Aug 2011
6 Sep 2011
6 Oct 2011
0.5
i
i
i
perrucci 2012/7/11 17:12 page 112 #134
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 113 #135
INFLATION-LINKED MARKETS
Figure 7.2 Ten-year US TIPS real yields and real GDP growth
4.5
4.0
3.5
3.0
2.5
%
2.0
1.5
1.0
10-year real yield
0.5
Q4 2009
Q4 2008
Q4 2007
Q4 2006
Q4 2005
Q4 2004
Q4 2003
Q4 2002
Q4 2001
Q4 2000
Q4 1999
Q4 1998
Q4 1997
i
i
i
perrucci 2012/7/11 17:12 page 114 #136
INFLATION-SENSITIVE ASSETS
and W T W max
(the superscript T indicates the transpose). Mathematical machinery aside, it is interesting to analyse how an investor might use such
an allocation process in practice. As an example, consider a retail
investor who is engineering a financial plan to ensure a safe and
comfortable retirement. To this end, they will initially need to estimate and commit to a sustainable saving and investment schedule,
with the objective being to have enough proceeds later in life to satisfy their consumption needs. These needs also have to be estimated
and, as mentioned before, a consumption level should be targeted
in real rather than nominal terms, given the uncertainty on inflation.
Based on these projections, the investor can infer the minimum portfolio real return that is required, ie, rmin . Then, depending on their
risk and volatility tolerance, they can solve for the optimal portfolio.
Of course, there is a possibility that this optimal portfolio might not
114
i
i
i
perrucci 2012/7/11 17:12 page 115 #137
INFLATION-LINKED MARKETS
US TIPS
160
140
autumn 2008
Flight to liquidity
120
100
Mar 2012
Mar 2011
Mar 2010
Mar 2009
Mar 2008
Mar 2007
Mar 2006
Mar 2005
Mar 2004
Mar 2003
Mar 2002
Mar 2001
Mar 2000
Mar 1999
Mar 1998
Mar 1997
80
exist, in which case the investor will need to save more, bear more
risk or adjust their projected retirement lifestyle.
In essence, we propose to focus on liability matching in real terms,
ie, funding future retirement consumption, instead of a simple asset
allocation framework. The real return constraint provides a baseline
investment objective, and in effect ensures that the investor is taking the minimum level of risk compatible with funding future consumption. Of course, this does not prevent them from being more
aggressive if their risk tolerance allows.
Clearly, in a liability-matching framework in real return space,
inflation-linked bonds become the true risk-free asset, as their return
volatility is limited (see our previous discussion on the linkage with
real economic growth), and their cash proceeds are indexed to inflation. This is why they contribute to a better overall portfolio when
added into the mix with other asset classes. In addition, in contrast to
an investment in traditional asset classes such as equities and nominal bonds, their real return profile is intrinsically more stable and is
not dependent on inflation forecasts, and thus their model outputs
are less sensitive to biases in historical estimates or regime shifts in
inflation going forward.
115
i
i
i
perrucci 2012/7/11 17:12 page 116 #138
INFLATION-SENSITIVE ASSETS
Average (%)
Standard deviation
Correlation
Number of observations
Monthly
nominal
returns
US
Treasury
US
TIPS
US
Treasury
US
TIPS
0.31
1.49
0.36
1.72
0.51
1.36
0.56
1.72
66
175
64
175
i
i
i
perrucci 2012/7/11 17:12 page 117 #139
INFLATION-LINKED MARKETS
4.4
(a)
4.3
4.2
4.1
Optimal blend: 50% Treasury and 50% TIPS portfolio
4.0
3.9
3.8
100% Treasury portfolio
3.7
4.8
6.8
5.0
5.2
5.4
5.6
5.8
Annualised volatility of real returns (%)
(b)
6.0
6.7
6.6
6.5
6.4
Optimal blend: 70% Treasury and 30% TIPS portfolio
6.3
6.2
100% Treasury portfolio
6.1
4.4
4.6
4.8
5.0
5.2
5.4
5.6
Annualised volatility of nominal returns (%)
5.8
6.0
Results for (a) real return space; (b) nominal return space.
Source: data from New Sky Capital, Barclays.
corrected and capital surpluses from savers and investors are put to
use by producers and issuers, respectively.
A secondary market is also active, and this is where investors
can rebalance and adjust their portfolios and issuers can get price
indications for where the market is trading. In the secondary market for inflation-linked bonds, liquidity is relatively good, albeit
117
i
i
3.0
1.5
3.0
0.5
i
18 Oct 2010
1 Nov 2010
15 Nov 2010
29 Nov 2010
13 Dec 2010
27 Dec 2010
10 Jan 2011
24 Jan 2011
2 Feb 2011
21 Feb 2011
7 Mar 2011
21 Mar 2011
4 Apr 2011
18 Apr 2011
2 May 2011
16 May 2011
30 May 2011
13 Jun 2011
27 Jun 2011
11 Jul 2011
25 Jul 2011
8 Aug 2011
22 Aug 2011
5 Sep 2011
19 Sep 2011
3 Oct 2011
18 Oct 2010
1 Nov 2010
15 Nov 2010
29 Nov 2010
13 Dec 2010
27 Dec 2010
10 Jan 2011
24 Jan 2011
2 Feb 2011
21 Feb 2011
7 Mar 2011
21 Mar 2011
4 Apr 2011
18 Apr 2011
2 May 2011
16 May 2011
30 May 2011
13 Jun 2011
27 Jun 2011
11 Jul 2011
25 Jul 2011
8 Aug 2011
22 Aug 2011
5 Sep 2011
19 Sep 2011
3 Oct 2011
i
i
(a)
2.5
(b)
2.5
INFLATION-SENSITIVE ASSETS
Figure 7.5 Real yield and inflation break-evens for selected US TIPS
TII Q5 Apr 15 real
TII Q5 Apr 15 BE
2.0
1.5
%
1.0
1.5
0.5
1.0
2.0
%
1.5
1.0
0.5
118
i
perrucci 2012/7/11 17:12 page 119 #141
INFLATION-LINKED MARKETS
(c)
2.5
2.0
% 1.5
1.0
TII 1.125 Feb 40 real
1.5
not nearly as good as in other larger asset classes, such as nominal government bonds. This opens the way for temporary price
dislocations, and thus opportunistic plays, where providing liquidity at the right time is greatly rewarded. Because of the very nature
of these opportunities, most liquidity providers are sophisticated
leveraged investors, such as banks proprietary desks and hedge
fund managers, who have their finger on the pulse of technical
flows, and can quickly position accordingly. Although their time
horizon can vary, it is usually short to medium term (typically a
few days or weeks), and thus complements other investors, such
as traditional asset managers, who tend to have a more fundamental long-term approach. Furthermore, while several non-leveraged
investors have a long-only bias (investing in real rates), these liquidity providers take positions on both the long and short sides of the
market, thus betting on either real rates or inflation break-evens.17
Because of the diversity in investors objectives, time horizons and
valuation approaches, the market might be pricing certain securities
or market variables efficiently, while at the same time pricing others
inefficiently.
119
i
i
i
perrucci 2012/7/11 17:12 page 120 #142
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 121 #143
INFLATION-LINKED MARKETS
i
i
i
perrucci 2012/7/11 17:12 page 122 #144
INFLATION-SENSITIVE ASSETS
right issue and at the right time) to add value to an active portfolio
by just achieving superior execution.
Other inefficiencies: market-on-close prices, new issue
auctions and index rebalancing
The influence of passive-indexed-investors in the US TIPS market
also creates other opportunities. To start, as benchmarks use marketon-close prices to calculate returns, many passive investors tend to
trade at, or near to, market close to better track their index. Indeed, it
is not uncommon to see the market richen near closing time (by half a
point or even more) purely owing to this technical effect, and to then
cheapen again the following day. This structural pattern has been
consistent over time, and offers opportunity for the active investor
to buy cheap bonds intra-day and sell them rich near the close of the
trading session.
In addition, new issue auctions and month-end rebalancing are
two calendar events that typically have a material technical effect on
the market. For example, it is not unusual for US TIPS to cheapen in
the weeks ahead of auction dates (there are eight US TIPS auctions
each year) in anticipation of supply, and then richening again in the
weeks following, as many passive investors are obliged to buy newly
issued benchmark-eligible securities. A similar mechanism is at play
when the benchmark index is rebalanced at the end of each month.
Although the specific issues that will be leaving or joining the index
are known in advance, passive investors, wishing to closely match
their benchmark price performance, tend to rebalance their portfolio
only close to the end of the month. This offers a structural mechanism
to add value for active managers who can anticipate buying/selling
activity and position accordingly.
Risk-adjusted inflation break-even rate
In most market conditions, inflation-linked bonds have lower price
volatility than the nominal bonds of similar maturities. This is
because, despite the fact that linkers have higher interest rate duration, real yields tend to be less volatile than nominal yields.18 This
implies that risk-constrained investors can hold a higher notional of
inflation-linked bonds than of notional bonds. Therefore, if we were
to compute the inflation break-even rate between a real bond and a
smaller risk-adjusted notional of the equal maturity nominal bond,
the result would be a risk-adjusted break-even rate smaller than the
122
i
i
i
perrucci 2012/7/11 17:12 page 123 #145
INFLATION-LINKED MARKETS
i
i
i
perrucci 2012/7/11 17:12 page 124 #146
INFLATION-SENSITIVE ASSETS
Corporate
issuer
Inflation-linked bond
Corporate
issuer
Nominal bond
Inflation-linked revenues
Pay inflation
in a swap
Receive fixed
(b)
Inflation swap
(a) Issuing an inflation-linked bond. (b) Issuing a nominal bond and paying inflation
in a swap.
Inflation-linked bond
Pay fixed
in a swap
Receive inflation
Corporate
issuer
Inflation swap
i
i
i
perrucci 2012/7/11 17:12 page 125 #147
INFLATION-LINKED MARKETS
i
i
i
perrucci 2012/7/11 17:12 page 126 #148
INFLATION-SENSITIVE ASSETS
Inflation
Inflation
payer
Inflation
receiver
Fixed rate K %
At maturity the inflation payer pays notional (index(T )/index(base) 1) and the
index receiver pays notional ((1 + K%)T 1).
R % + inflation
Inflation
payer
Inflation
receiver
Libor (quarterly)
Each year Y the inflation player pays notional (R% + max[index(Y )/index(Y
1) 1, 0]). The inflation receiver pays quarterly notional Libor day count.
Inflation linked
Inflation
payer
Inflation
receiver
Libor (quarterly)
i
i
i
perrucci 2012/7/11 17:12 page 127 #149
INFLATION-LINKED MARKETS
for off-balance-sheet inflation exposure (ie, for receiving inflationlinked cashflows in a swap) has often, albeit not always, outstripped
supply, resulting in a premium offered to inflation payers relative to
the cash market (we hinted as to this effect when comparing parts
(a) and (b) of Figure 7.6).
The simplest and most liquid inflation swap is the zero-coupon
swap, with only one cashflow exchanged at maturity. Specifically, at
maturity T, the inflation payer will pay the realised inflation over
the life of the contract, in exchange for a previously agreed fixed rate
K%. This is shown schematically in Figure 7.8.
In each market, the price index underlying the swap is typically
the same as the corresponding bond market (eg, CPI (all urban consumers, non-seasonally adjusted) for the US), while rules for the
calculation of the relevant daily price index, in particular the length
of the lag and whether or not interpolation is used, can vary quite a
lot.20 As for settlement, the inflation swap market typically follows
nominal swap market conventions (eg, t + 2 in the US and EU, t+ in
the UK), and thus usually differs from the bond market.
On each trading day, zero-coupon inflation swap quotes for
annual tenors (typically up to 10 years, although sparser quotes
often extend to longer maturities) are readily available (for example, on Bloomberg and other brokers screens). From these inflation
swap quotes, it is straightforward to derive a discrete (annual intervals) curve of forward index price levels and implied inflation rates.
To construct a continuous (in this context, monthly) curve,21 two
additional tools are required:
a method for interpolation (linear, cubic spline, etc) between
The inflation swap curve thus derived provides the base from which
all other more complex swaps can be priced.
Another common swap is the additive, or year-over-year, inflation swap. In this structure, one party pays a fixed coupon plus
inflation (typically floored at zero) annually in exchange for fixed or
variable (Libor) rate payments. As mentioned in the previous section, this structure (Figure 7.7) has developed to match similarly
additive corporate inflation-linked notes issuance (common in the
European retail market in particular), where there is no accretion on
127
i
i
i
perrucci 2012/7/11 17:12 page 128 #150
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 129 #151
INFLATION-LINKED MARKETS
Bond
Bond cashflows
ASW
buyer
Bond cashflows
ASW
seller
Spread
Down
N/A
N/A
N/A
Up
Down
N/A
N/A
Down
N/A
N/A
N/A
N/A
Down
N/A
N/A
N/A
N/A
Up
Up
Up
Down
Down
Down
Up
Up
Up
Up
Up
Up
Slightly
down because of the small duration from the ASW fixed spread.
up because of the small duration from the ASW fixed spread.
The market movements shown are essentially partial derivatives. The four
variables are the nominal Treasury yield corresponding to the bond maturity, the swap yield for the same maturity, the bond liquidity spread and the
bond credit spread. For example, in column (a) the Treasury yield goes
up, while the other three variables are fixed (unchanged swap yield means
that the swap spread narrows). In column (b), the swap yield goes up, but
the Treasury yield, the bond liquidity spread and the bond credit spread
are unchanged.
Slightly
i
i
i
perrucci 2012/7/11 17:12 page 130 #152
INFLATION-SENSITIVE ASSETS
US
TIPS
US TIPS cashflows
ASW
buyer
US TIPS cashflows
ASW
seller
(greater) amount than the price of the bond, so will have to pay
the premium to (or receive the discount from) the ASW buyer up
front. In either case, the ASW spread can be determined by setting
the net present value of the swap cashflows equal to zero, but the
specific formulas, and the resulting ASW spreads, differ depending
on whether the swap is quoted on a proceeds or parpar format.
As an example, suppose that the bond has a full (dirty) price P,
pays annual coupons equal to Cj and has redemption amount B at
maturity T; i is the swap curve discounting factor for maturity ti
(ie, the present value of a US dollar paid at time ti ); li is the Libor
paid semi-annually at time ti (and set at time ti1/2 ); i is the day
count to be applied to floating-rate payments at time ti . Then, for a
proceeds ASW, where the swap notional is equal to the bond dirty
price P, the ASW spread calculated as
2T
T
P
i/2 i/2 (li/2 + ASWproceeds ) + (P B)T
j Cj = 0
100 k =1
j =1
notional exchange
Libor cashflows
bond coupons
2T
k =1
+ (100 B)T
T
j C j = 0
j=1
k =1
130
i
i
i
perrucci 2012/7/11 17:12 page 131 #153
INFLATION-LINKED MARKETS
ASWproceeds =
ASWparpar =
j=1 j Cj + BT P
2T
1
k=1 i/2 i/2
100 P
P
ASWproceeds
100
from which we can see that the net present value of the ASWproceeds
spread payments compensates for the difference arising from discounting the bond cashflows using the swap curve versus the relevant bond curve. Also, note that the parpar and proceeds ASW
spreads are the same only for bonds trading at par. Therefore, it is
not straightforward to compare bonds using ASW spreads, and the
zero volatility spread (ZV), which is the continuously compounded
spread on top of swap rates needed to recover the price of the bond,
is often used instead
P = BT exp( ZV tT ) +
T
j Cj exp( ZV tj ) = 0
j =1
When the asset underlying the swap is an inflation-linked bond (Figure 7.12 for an example depicting US TIPS as the underlying asset
bond), the analysis is complicated by the fact that we need to project
nominal inflation-linked payments into the future.
To this end, inflation index values implied by zero-coupon swaps
are used to project notional accretion on the inflation-linked bonds,
ie, the index factor ratios IR(J ), from trade date to maturity. One
can then use the formulas derived above simply by explicitly introducing index factors and the inflation floor at maturity. Specifically
(assuming the day count factor for the annual coupon payment is 1
for simplicity of notation)
CJ = C IR (J ),
B = 100 max[IR(T ), 1]
i
i
i
perrucci 2012/7/11 17:12 page 132 #154
INFLATION-SENSITIVE ASSETS
Table 7.5 The effect of interest rate, spread and inflation break-even
changes
Interest rate
Market value of
bond and swap
ASW buyer
Inflation-linker
Spread
Inflation
Tsy Swap
swap
yield yield Liquidity Credit
BE
up
up
up
up
up
(a)
(b)
(c)
(d)
(e)
Down
N/A
Down
Down
Up
Pay linker
cashflows
N/A
Up
N/A
N/A
Down
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Receive linker
cashflows
N/A
Down
N/A
N/A
Up
Parpar ASW
Up
Down
Up
Up
No
change
Proceeds ASW
Up
Down
Up
Up
No
change
Up
Down
Up
Up
Down
ASW seller
Pay Libor + AWS
Slightly
down because of the small duration from the ASW fixed spread.
up because of the small duration from the ASW fixed spread.
The sensitivities shown are essentially partial derivatives, where one variable is moved and the remaining four are held fixed. For example, in column
(a) the Treasury nominal yield goes up, but the inflation break-even does
not move, and so the other spreads and thus the linker price go down (in
other words, the real yield moves in lockstep with the nominal yield, thus
driving down the linker price). In column (e), the inflation break-even goes
up, but the nominal Treasury yield is held fixed, implying that the real yield
goes down, thus increasing the price of the linker. The ASW spread does
not change with a change in inflation break-even, as the swap fixed-leg
change in market value is picked up either by the upfront payment (in a
parpar swap) or by a change in swap notional (for a proceeds swap).
Slightly
typically also contains a liquidity component, and market inflation expectations are better measured by zero-coupon (ZC) swap
inflation break-evens. Convexity considerations aside (the cash inflation break-even on a coupon bond will be different from a ZC swap
break-even even in the absence of a bond liquidity premium), we
132
i
i
i
perrucci 2012/7/11 17:12 page 133 #155
INFLATION-LINKED MARKETS
can write
real yield r n BEcash
n BEswap + liquidity spread
i
i
i
perrucci 2012/7/11 17:12 page 134 #156
INFLATION-SENSITIVE ASSETS
Obviously, we are referring to the modern inflation-linked market. Historically, the indexation
of government debt has deeper roots that can be traced back to bills linked to the price of
silver issued by the Massachusetts Bay Colony in 1742.
Note that EU countries have at times issued linkers indexed to either their domestic inflation
index or the Euro Harmonized Index of Consumer Prices (HICP).
There has been mixed empirical evidence on the inflation risk premium, as often the liquidity premium between nominal and inflation-linked bonds acts in the opposite direction,
especially in newly established markets. However, it is plausible that a positive inflation risk
premium exists more often than not in liquid markets, especially for long tenors, where the
insurance benefit is more important.
The CPI released in month n is released about two weeks into the following month.
The quoted price is expressed as an integer (103) plus ticks (4, one tick being 1/32) and half a
tick (+ = 1/64). Therefore, 103 4+ translates into 103 + (4/32) + (0.5/32) = 103.140625.
Fishers equation neglects the risk premium, which is a function of the correlation between
real rates and inflation.
This is often, but not always (contrast with what happened in 20089) a reasonable approximation for long-maturity linkers or for when the floor is far out-of-the-money (old versus
new issue linkers). Alternatively, option-adjusted break-even measures can be developed by
accounting for the value of the deflation floor explicitly.
However, the market has become increasingly efficient when it comes to pricing seasonality.
10 As most corporates do not issue inflation bonds, this will be a nominal rate adjusted by ex
ante expected inflation, or ex post realised inflation rate.
11 The Greek debt crisis started in 2009 and is ongoing at the time of writing; this affected other
developed market countries in Europe and elsewhere (eg, US, Japan, Spain, Ireland, Portugal,
Italy, UK) with unsustainably high levels of debt.
134
i
i
i
perrucci 2012/7/11 17:12 page 135 #157
INFLATION-LINKED MARKETS
12 Note that governments have traditionally been considered more creditworthy than domestic
companies, at least on debt denominated in local currency, because of the ability to print
money, thus inflating debt away and devaluing the currency. This may no longer be the case
for the Eurozone countries, which have forgone this option by adopting a common currency.
13 Potential GDP growth is not directly observable, but estimates can be derived by empirical
study of equilibrium relations as well as macroeconomic considerations.
14 To analyse the link between real rates and economic growth, New Sky Capital uses a proprietary blend of realised and potential GDP (as published by the US Federal Reserve; see
http://research.stlouisfed.org).
15 Note that the Barclays US Treasury Index duration is smaller (about five years on average)
than the Barclays US TIPS Index duration (about eight years on average). It is also plausible
that some of the outperformance of US TIPS over nominal Treasuries might be due to a
progressive narrowing of the liquidity/novelty premium on inflation-linked bonds.
16 Here and in the figures, by optimal portfolio we mean the portfolio with highest expected
return per unit risk.
17 This being said, many long-only money managers also look at inflation-linked bonds as a
substitute to nominal government securities in their benchmarks. Thus, these investors also
express an implicit view on break-evens.
18 This is a consequence of the positive correlation between real rates, nominal rates and inflation. Typically, in a bond rally (sell-off), both real rates and market inflation expectations will
decrease (increase); thus, nominal rates will generally be more volatile than real rates (exceptions occur, especially at the front end of the real curve, which is more sensitive to inflation
surprises).
19 Differences in haircuts and repo rates between nominal bonds and linkers (usually more
vexing for the latter) are second-order effects for the purpose of this calculation.
20 For the US swap market the lag and interpolation formula are the same as those used for US
TIPS, but for the UK and Eurozone swap market there is a lag but no daily interpolation in
the swap reference price index.
21 Special care and additional instruments such as inflation futures are typically used to construct
the short end of the curve (with tenor less than one year), but a discussion is beyond the scope
of this section. Note that inflation price index seasonality translates into greater amplitude
of seasonality for inflation swap rates in short versus long tenors (as the same price index
seasonality is damped in long-term swap rates).
22 This spread can be positive or negative (depending on the relative credit and liquidity risks
of the underlying bond and the swap market).
23 A synthetic floater is one obtained by combining several underlying transactions, whose
cashflows effectively combine.
24 In addition, both ASW parties assume counterparty credit risk.
25 What is meant here is that if both Treasury yield and swap yield move in lockstep, the market
value change in the bond and the fixed leg of swap offset. Technically, there is a residual
interest rate risk resulting from different discounting curve on the asset (bond) and swap side
(ie, different durations even if all discounting curves involved move in parallel).
REFERENCES
135
i
i
i
perrucci 2012/7/11 17:12 page 136 #158
INFLATION-SENSITIVE ASSETS
Markowitz, H., 1959, Portfolio Selection: Efficient Diversification of Investments (New York:
John Wiley & Sons).
Perrucci, S., 2010, Inflation-Sensitive Assets: Portfolio Benefits and Opportunities,
Report, URL: http://www.newskycapital.com.
Perrucci, S., 2011 Efficient Frontier and Optimal Portfolios in Real vs. Nominal Space,
Report, URL: http://www.newskycapital.com.
136
i
i
i
perrucci 2012/7/11 17:12 page 137 #159
i
i
i
perrucci 2012/7/11 17:12 page 138 #160
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 139 #161
Pension fund
cashflows
( thousands)
Member retires
50
25
0
Years
2.
Inflows
Pension fund
cashflows
( millions)
Member dies
1.
Outflows
Inflation
75
50
25
0
Years
During the accumulation period (inflows) the pension contributions are implicitly
linked to wage inflation. During the payment period (outflows), pension annuities
may be explicitly linked to consumer price inflation. 1. In the UK the annuities are
linked to the RPI. In the Netherlands the annuities can be linked to the CPI. 2. The
first annuity is generally proportional to the final contribution.
At this point, we shall review the reasons why pension funds are
exposed to the risk of unexpected high inflation in the future. Simply
put, a pension fund cashflows stream can be split into two periods,
as shown in Figure 8.1:
1. during the accumulation (cash inflows into the pension fund)
period, pension scheme members contribute funds via regular
instalments during their active working life;
2. when members reach retirement age, the distribution (cash
outflows from the pension funds) period begins, and the retiree
receives a pension annuity until their death.
The inflation sensitivity of these cashflows is a function of actuarial
projections and accounting valuations. The latter are in turn influenced by pension regulations, which differ from country to country
and evolve over time. But the crucial point is to understand how
price levels affect the initial amount of benefits payable and their
growth over time.
139
i
i
i
perrucci 2012/7/11 17:12 page 140 #162
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 141 #163
i
i
i
perrucci 2012/7/11 17:12 page 142 #164
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 143 #165
Inflation
Inflation
Pension
funds
Banks
Swap break-even
inflation
Swap break-even
inflation
i
i
i
perrucci 2012/7/11 17:12 page 144 #166
INFLATION-SENSITIVE ASSETS
2%+ inflation
+ inflation YoY floor
Note
issuers
Bank
Floating rate
Note
buyer
Cash at inception
i
i
i
perrucci 2012/7/11 17:12 page 145 #167
Figure 8.4 ZCIS quotes as of July 26, 2011, for the French and the
EMU HICPs (Bloomberg tickers: FRCPXTOB and CPTFEMU)
Counterpart
CPI(10Y 3M)
CPI(today 3M)
i
i
i
Source
US
Japan
UK
Australia
Ex-tobacco
HICP
Unrevised
NSA
Ex-tobacco
CPI
Unrevised
NSA
CPI
urban
Unrevised
NSA
Ex-fresh
food CPI
Unrevised
NSA
RPI All
items
Unrevised
NSA
RPI All
items
Unrevised
NSA
Eurostat
INSEE
BLS
MPM
ONS
ABS
Bloomberg
CPTFEMU
<Index>
FRCPXTOB
<Index>
CPURNSA
<Index>
JCPNGENF
<Index>
UKRPI
<Index>
AUCPI
<Index>
Frequency
Monthly
Monthly
Monthly
Monthly
Monthly
Quarterly
Straight
France
Straight
Daily
Daily
Daily
Straight
3M
2M3M
2M3M
2M3M
2M
1Q
High
Medium
High
Low
Medium
Medium
swap index changes once a month. The swap index is calculated daily as a linear interpolation between two consecutive inflation indexes.
CPI
EMU
INFLATION-SENSITIVE ASSETS
146
Table 8.1 Different market conventions for inflation swaps
i
i
i
perrucci 2012/7/11 17:12 page 147 #169
CPI(t lag)
1
CPI(t 1Y lag)
Part of the forward CPI curve can be derived directly from the zerocoupon swap rates (ie, for the maturities for which ZCIS are quoted).
If we take the example of the previous section, we can calculate the
10Y forward CPI with a three-month lag from the quoted price of the
10Y ZCIS 2.285% as follows (note that the initial index CPI(today
3M) is a known quantity, which is worth 112.75 in the example)
CPI(10Y 10M) = CPI(t 3M)(1 + 2.285%)10
= 112.75(1 + 2. 285%)10
= 141. 33
However, these yearly points are incomplete and need to be extended to monthly or daily points in order to recover the whole
pricing curve, a topic covered in the next subsection.
Convexity adjustments in YoY inflation swaps
Prices of YoY swaps are not really observable, but can be derived
from ZCIS prices, although appropriate convexity adjustments are
required. These adjustments come from the fact that the YoY inflation
swaps (YYISs) can only be partially hedged with a portfolio of ZCISs,
as we shall see later. These mismatches are quantified by using specific stochastic models. Each trader has their own model and calibration techniques. As a result, the adjustments, and therefore the
pricing of YoY swaps, can differ materially.
Incidentally, at-the-money (ATM) YoY inflation option prices also
differ, given the lack of consensus on the pricing of the underlying.
Indeed, the strikes of YoY inflation options are expressed in absolute
levels (typically 2%, 1%, 0%, 1%, 2%, 3%, 4%, 5%) and not as a
spread relative to the current ATM level, as is the case for most
options on nominal yields (for example, ATM +50 basis points).
An intuitive way to understand the convexity adjustment is to
analyse how to hedge a YYISs with ZCISs, which trade in the interbanking market. As an illustration, we shall consider the 10Y cashflows of a YYIS, where the inflation leg is based on the yearly inflation
147
i
i
i
perrucci 2012/7/11 17:12 page 148 #170
INFLATION-SENSITIVE ASSETS
10Y
2.2%
I10 / I9
1.94%
9Y
10Y
I9 / I0
I10 / I0
1.97%
Reinvestment rate r9
2.0
300
1.5
200
Millions
1.0
100
0.5
0
0.5
1.0
1.5
2.0
1.5
Portfolio
Hedge
Long portfolio,
short hedge
1.0
0.5
0
0.5
1.0
Change in 9Y and 10Y BEI (parallel shift, %)
100
200
300
1.5
Figure 8.7 Impact of changes in BEI9 and BEI10 (parallel shift) on the
global P/L for a notional of 100M: positive convexity
rate, which will accrue between 9Y and 10Y; we denote this stochastic cashflow by YoY9,10 . The question is: what is todays value of this
cashflow or the break even inflation rate (denoted by BEI9Y,10Y )?
1. The bank portfolio to be hedged consists of a short position
in a YoY9,10 swap. In other words, in 10 years time, the bank
pays I10 /I9 1 (where I9 and I10 are the stochastic 9Y and 10Y
forward inflation indices) and receives BEI9Y,10Y (a level that is
defined today). Clearly, the profit or loss (P/L) of this position
is proportional to the change in I10 /I9 .
2. The hedge consists of two ZC swaps.
(a) Short 9Y ZCIS: the bank pays I9 /I0 and receives todays
BEI9 in nine years.
148
i
i
i
perrucci 2012/7/11 17:12 page 149 #171
Thousands
100
50
0
50
Portfolio
Hedge
Long portfolio,
short hedge
100
150
200
1.5
1.0
0
0.5
1.0
0.5
Change in 9Y and 10Y BEI (parallel shift, %)
1.5
3.1
2.9
2.7
2.5
2.3
Linear
YoY forward cubic spline
YoY forward linear
Cubic spline
2.2
1.9
1.7
0
10
15
20 25 30
Maturity (years)
35
40
45
50
(b) Long 10Y ZCIS: the bank receives I10 /I0 and pays todays
BEI10 in ten years, where I0 is todays known inflation
index.
The P/L is proportional to the change in the expected value of both
I10 and I9 , and the one-year reinvestment rate r9 of the 9Y net cashflow
into 10Y (Figure 8.6).
The two main sources of convexity are the following.
1. Figure 8.7: the hedge P/L is proportional to the change in the
expected value of I9 and I10 , whereas the YoY swap P/L is not (it
149
i
i
i
perrucci 2012/7/11 17:12 page 150 #172
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 151 #173
115
110
11
13
15
17
19
those tenors, and there is no liquid real rate product for short maturities either. However, within the one-year horizon, economists forecasts are fairly accurate and can provide some guidance in building
the short end of the inflation curve.
Another issue has to do with seasonality, as most indexes used in
the inflation market are non-seasonally adjusted (NSA). In addition,
seasonality information can be derived partly from inflation-linked
bonds, but with limited accuracy. Last, seasonality factors change
over time and there is no accepted market standard as to how they
should be computed.
In Table 8.3, we have chosen to use the latest European Central
Bank seasonality factors (European Central Bank 2000) in the sixth
column as the best predictors for future seasonality.
Table 8.3 shows an example of calculations where economists
forecasts (third column) are used as a guide to build the short end
of the curve (fifth column) until December 2012. From January 2013
onwards, the monthly CPI curve is calculated from the yearly forward CPI (4th column) implied by the ZCIS prices. The issue is how
we can build monthly CPI from yearly CPI, which are non-seasonally
adjusted (NSA). As Figure 8.10 shows, the seasonality creates an
oscillating shape, which prevents interpolation. The solution is first
to make the data seasonally adjusted (SA), by dividing the fifth column by the sixth column to obtain the seventh column (labelled as
SA). Then in the eighth column we can interpolate the data to obtain
the monthly SA CPI. In the ninth column we convert the monthly
SA CPI back into NSA CPI.
151
i
i
i
0Y
1Y
2Y
3Y
4Y
5Y
6Y
7Y
8Y
9Y
10Y
11Y
12Y
13Y
14Y
15Y
Maturity
06/08/2012
05/08/2013
04/08/2014
04/08/2015
04/08/2016
04/08/2017
06/08/2018
05/08/2019
04/08/2020
04/08/2021
04/08/2022
04/08/2023
05/08/2024
04/08/2025
04/08/2026
1.750
1.770
1.857
1.913
1.975
2.040
2.070
2.107
2.146
2.173
2.201
2.242
1.872
1.886
1.941
1.969
2.033
2.102
2.140
2.173
2.183
2.210
2.260
2.310
2.314
2.319
2.323
1.872
1.886
1.941
1.969
2.033
2.102
2.140
2.173
2.182
2.210
2.262
2.310
2.330
2.329
2.323
HICP
Date
05/2011
05/2012
05/2013
05/2014
05/2015
05/2016
05/2017
05/2018
05/2019
05/2020
05/2021
05/2022
05/2023
05/2024
05/2025
05/2026
Fwd HICP
YoY fwd
Fwd HICP
Linear Cubic spline Linear Cubic spline Cubic spline
112.74
114.85
117.03
119.43
121.89
124.68
127.73
130.75
133.90
136.92
140.29
144.16
148.28
151.79
155.40
159.10
112.74
114.85
117.03
119.43
121.89
124.68
127.73
130.75
133.90
136.92
140.29
144.19
148.28
152.09
155.62
159.10
1.87
1.90
2.05
2.05
2.29
2.45
2.37
2.40
2.26
2.46
2.76
2.86
2.37
2.38
2.38
1.87
1.90
2.05
2.05
2.29
2.45
2.37
2.40
2.26
2.46
2.78
2.84
2.57
2.32
2.24
112.74
114.85
117.03
119.43
121.89
124.68
127.73
130.75
133.90
136.92
140.29
144.19
148.28
152.09
155.62
159.10
Tenor
Interpolated curve
Market mid
ICPI
Linear Cubic spline
INFLATION-SENSITIVE ASSETS
152
Table 8.2 ZCIS quotes, forward index levels and YoY inflation rates
Trading date: August 3, 2011. Settlement date: August 4, 2011. Yearly tenors.
i
i
i
Table 8.2 (Cont.)
04/08/2027
04/08/2028
06/08/2029
05/08/2030
04/08/2031
04/08/2032
04/08/2033
04/08/2034
06/08/2035
04/08/2036
04/08/2037
04/08/2038
04/08/2039
06/08/2040
05/08/2041
2.283
2.298
2.377
2.332
2.341
2.351
2.360
2.369
2.371
2.374
2.376
2.379
2.381
2.395
2.409
2.424
2.438
2.452
2.324
2.333
2.346
2.359
2.369
2.374
2.376
2.376
2.377
2.381
2.390
2.404
2.420
2.436
2.452
05/2027
05/2028
05/2029
05/2030
05/2031
05/2032
05/2033
05/2034
05/2035
05/2036
05/2037
05/2038
05/2039
05/2040
05/2041
Fwd HICP
YoY fwd
Fwd HICP
Linear Cubic spline Linear Cubic spline Cubic spline
163.03
167.09
171.28
175.61
180.07
184.43
188.90
193.49
198.20
203.03
208.62
214.41
220.43
226.69
233.18
Trading date: August 3, 2011. Settlement date: August 4, 2011. Yearly tenors.
162.83
166.86
171.13
175.57
180.07
184.53
188.97
193.46
198.11
203.03
208.36
214.10
220.20
226.59
233.18
2.47
2.49
2.51
2.53
2.54
2.42
2.42
2.43
2.43
2.44
2.75
2.78
2.81
2.84
2.86
2.35
2.47
2.56
2.59
2.57
2.48
2.41
2.38
2.40
2.49
2.63
2.75
2.85
2.90
2.91
162.83
166.86
171.13
175.57
180.07
184.53
188.97
193.46
198.11
203.03
208.36
214.10
220.20
226.59
233.18
153
16Y
17Y
18Y
19Y
20Y
21Y
22Y
23Y
24Y
25Y
26Y
27Y
28Y
29Y
30Y
Maturity
HICP
Date
Tenor
Interpolated curve
Market
mid Linear Cubic spline
i
i
i
May 10
Jun 10
Jul 10
Aug 10
Sep 10
Oct 10
Nov 10
Dec 10
Jan 11
Feb 11
Mar 11
Apr 11
May 11
Jun 11
Jul 11
Aug 11
Sep 11
Oct 11
Nov 11
Dec 11
109.71
109.70
109.32
109.54
109.77
110.15
110.27
110.93
110.11
110.57
112.11
112.75
112.74
112.75
Fwd
CPI
112.74
112.25
112.71
113.24
113.67
113.74
114.19
NSA
Seasonals
SA Fwd
SA Fwd
NSA
NSA SA
Fwd CPI
2010
CPI
CPI
MoM
MoM MoM
Forecasts
ECB
(incomplete) Linear Fwd CPI (%) (%)
109.71
109.70
109.32
109.54
109.77
110.15
110.27
110.93
110.11
110.57
112.11
112.75
112.74
112.75
112.25
112.71
113.24
113.67
113.74
114.19
1.0044
1.0044
0.9996
0.9999
0.9998
1.0011
0.9998
1.0005
0.9921
0.9939
1.0007
1.0040
1.0044
1.0044
0.9996
0.9999
0.9998
1.0011
0.9998
1.0005
109.23
109.22
109.36
109.56
109.79
110.02
110.30
110.87
110.99
111.25
112.03
112.30
112.25
112.26
112.30
112.73
113.26
113.54
113.77
114.13
109.23
109.22
109.36
109.56
109.79
110.02
110.30
110.87
110.99
111.25
112.03
112.30
112.25
112.26
112.30
112.73
113.26
113.54
113.77
114.13
109.71
109.70 0.08 0.1
109.32 1.55 4.1
109.54 2.12 2.4
109.77 2.63 2.5
110.15 2.57 4.2
110.27 3.01 1.3
110.93 6.43 7.4
110.11 1.28 8.5
110.57 2.86 5.1
112.11 8.76 18.1
112.75 2.94 7.1
112.74 0.59 0.1
112.75 0.14 0.1
112.25 0.37 5.2
112.71 4.70 5.0
113.24 5.88 5.8
113.67 2.98 4.7
113.74 2.43 0.7
114.19 3.89 4.9
Tenor
CPI
month (publ.) Forecasts
INFLATION-SENSITIVE ASSETS
154
Table 8.3 Building a granular forward inflation curve
i
i
i
Table 8.3 (Cont.)
Tenor
CPI
month (publ.) Forecasts
114.85
117.03
113.08
113.38
114.65
115.05
115.09
115.15
114.49
114.73
115.16
115.43
115.38
115.70
117.03
0.9921
0.9939
1.0007
1.0040
1.0044
1.0044
0.9996
0.9999
0.9998
1.0011
0.9998
1.0005
0.9921
0.9939
1.0007
1.0040
1.0044
1.0044
0.9996
0.9999
0.9998
1.0011
0.9998
113.98
114.08
114.57
114.59
114.59
114.65
114.54
114.75
115.18
115.30
115.41
115.64
116.52
113.98
114.08
114.57
114.59
114.59
114.65
114.54
114.75
115.18
115.30
115.41
115.64
115.82
116.00
116.17
116.35
116.52
116.72
116.92
117.12
117.33
117.52
117.73
113.08 1.5411.1
113.38 1.00 3.2
114.65 5.30 14.3
115.05 0.25 4.3
115.09 0.07 0.4
115.15 0.66 0.6
114.49 1.18 6.7
114.73 2.22 2.5
115.16 4.68 4.6
115.43 1.20 2.9
115.38 1.15 0.5
115.70 2.43 3.4
114.90 1.90 8.0
115.29 1.89 4.1
116.25 1.71 10.4
116.81 1.89 6.0
117.03 1.82 2.3
117.23 2.11 2.1
116.87 2.04 3.6
117.11 2.10 2.4
117.30 2.10 2.0
117.66 2.03 3.7
117.70 2.09 0.4
113.08
113.38
114.65
115.05
115.09
115.15
114.49
114.73
115.16
115.43
115.38
115.70
NSA
Seasonals
SA Fwd
SA Fwd
NSA
NSA SA
Fwd CPI
2010
CPI
CPI
MoM
MoM MoM
Forecasts
ECB
(incomplete) Linear Fwd CPI (%) (%)
155
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
Apr 13
May 13
Jun 13
Jul 13
Aug 13
Sep 13
Oct 13
Nov 13
Fwd
CPI
i
i
i
perrucci 2012/7/11 17:12 page 156 #178
INFLATION-SENSITIVE ASSETS
From this curve, the inflation cashflows of most vanilla inflation products can be calculated and priced. However, more complex cashflows, which involve non-linear payouts or are based on
a forward starting index, require a full distribution and correlation
structure. Some of this information can be extracted from inflation
options, the topic of the next section.
INFLATION OPTIONS: A HISTORICAL PERSPECTIVE
This section will focus on options, in particular zero-coupon and
YoY inflation caps and floors, which are the most liquid inflation
options. We shall not cover more exotic instruments such as options
on real yields, inflation swaption or hybrid inflation options, which
would add too many dimensions to fit within one chapter. Readers
interested in a more complete analysis can refer to Bnaben and
Tabardel (2008) and Brigo and Mercurio (2006).
Floors in inflation-linked government bonds
When the US and France issued their first inflation-linked bonds in
the late 1990s, they embedded an inflation floor for the principal
payment at maturity. As some large institutional investors can only
invest in bonds that pay a guaranteed principal amount at maturity,
but cumulated inflation can in principle be negative, these issuers
added a principal protection in the form of an inflation floor. In other
words, at maturity T, the investor in the inflation-linked bond issued
at t = 0 receives
I (T )
, 100% notional
max
I (0)
where I (t) is the inflation index at the maturity of the linker and I (0)
is the inflation index underlying the linker. In practice, these deflation floors were given little attention, as their value was perceived
to be negligible, because most countries were experiencing inflation
rates within the range targeted by their monetary policies, and a
deflation episode would have to persist for a long time to affect the
redemption floor of long-maturity linkers.
In fact, both traders and investors regularly ignored these options
when valuing the linkers,3 and in the rare cases where they tried
to account for them explicitly, the models used were typically calibrated on historical data (when inflation was positive and stable),
so the risk was underestimated and not managed correctly.
156
i
i
i
perrucci 2012/7/11 17:12 page 157 #179
20,000
15,000
10,000
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
5,000
1988
25,000
Year of issuance
Source: Bloomberg, New Sky Capital, BNP Paribas.
The complacency from the world of practitioners did not prevent the academic world from working on a modelling framework
for valuing these options. For example, Robert Jarrow and Yildiray
Yildirim (2003) published on this topic, and provided a mathematical
formulation of how to the price the inflation floor.
The structured inflation notes market and YoY inflation options
Inflation trading desks started to look carefully at such models gradually, as the option risks on their books grew, which happened
hand in hand with the development of the inflation-structured notes
157
i
i
i
perrucci 2012/7/11 17:12 page 158 #180
INFLATION-SENSITIVE ASSETS
market. As mentioned before, strong retail demand developed (particularly in Italy) for inflation-protected structured notes. To meet
this demand, investment banks helped issuers to structure inflationlinked notes, with a typical maturity of around five years and an
annual coupon, comprising a fixed rate (say, 2% for our illustration) plus the realised year-on-year inflation. As a coupon cannot be
negative, the coupon paid at t was
max[2% + YoY(t), 0]
In other words, the annual coupons had an embedded inflation floor
at 2%. The issuers of such notes did not carry the inflation risk,
which was transferred to the trading books of the investment banks
via a swap. In other words, the investment banks were paying this
floored inflation to the issuers. In the absence of an inflation option
market, these floors were priced based on historical data, which
generally covered a period a relatively stable positive inflation. As
a result, the cost of such floors, as well as their risk, was underestimated. Indeed, many banks kept these short floor positions partially unhedged, with only limited reserves set aside to cover the
event of an increase in value (a loss on the short position). Over
time, the inflation-structured notes market grew significantly (Figure 8.11), due in part to the rise in commodity prices and its impact
on inflation.
These products became particularly popular with retail investors
in Europe and in the US. As shown in Figure 8.11, the peak of issuance
corresponds to 2007 (a year during which oil prices doubled) and the
first half of 2008 (when oil prices increased by about 50%). Because
of this large amount of issuance, banks accumulated large shortoption positions, mainly in YoY inflation floors struck at 0%. As the
risk grew, banks turned to other banks and brokers to offset their
exposure, which then gave birth to the interbank option market.
This is the main reason why the YoY format remains the most common structure in the inflation option market, in contrast to the swap
market, where the zero-coupon structure (similar to the cash bond)
tends to prevail.
The hard awakening of risk management in inflation option
books
The inter-broker market had an important consequence: it provided
market prices for inflation options. As a result, the positions in banks
158
i
i
i
perrucci 2012/7/11 17:12 page 159 #181
Figure 8.12 Change in delta while BEI decreases for 1M YoY floor
3
0
100
200
300
400
500
600
700
books started being marked-to-market, and the banks started hedging their trading book more actively. However, all banks had similar positions: they were short YoY floors and, consequently, they
could not really cover their exposures in the interbank market. That
led most banks to partially hedge these options, more precisely to
delta hedge the options; they sold a quantity of the underlying
inflation swap (by paying inflation and receiving a fixed rate), such
that the option price was immunised against changes in the price of
the underlying swap. Like in any option the delta hedge ratio varies
with the proximity to the strike price. Most of the floor levels were
around 0%, therefore the closer the BEI was to 0%, the more inflation
swap the banks had to sell to hedge the delta of the option.
As long as the BEI levels were stable at around 2% or 3%, this
hedging strategy worked well, with the banks pocketing the carry
of these options. Indeed, the issuance of YoY structured notes kept
rising until 2008. After the collapse of Lehman Brothers, the liquidity
crisis forced the trading books to sell most of their inflation-linked
bonds, causing BEI levels to collapse to unprecedented levels. For
example, in the US, the market was pricing deflation over a 10-year
horizon! The drop in BEI levels was also exacerbated by the hedging of these floors. In fact, as the BEI started decreasing and got
closer to the 0% strike, inflation option desks had to adjust their
delta hedge by selling more inflation swaps (Figure 8.12). This put
further pressure on BEIs, which in turn forced the inflation option
books to sell more swaps, and so on. This vicious circle caused huge
159
i
i
i
perrucci 2012/7/11 17:12 page 160 #182
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 161 #183
Figure 8.13 The difference between US bond and swap 10Y BEIs from
2006 to 2012
180
160
Basis points
140
120
100
80
60
40
20
0
Mar
2006
Mar
2007
Mar
2008
Mar
2009
Mar
2010
Mar
2011
Mar
2012
Figure 8.14 How selling the asset swap partly hedges the short
pension fund inflation swaps and the short YoY floors
Inflation
bond
issuers
1.
Floating
rate
ASW
buyer
Inflation
Pension
funds
Banks
2.
Fixed
rate
1. Debt: real cashflows plus inflation plus redemption inflation floor. 2. Swap: real
cashflows plus inflation plus inflation floor.
What is the connection between the asset swap market and the
option market? In fact, by selling asset swaps the banks bought some
zero-coupon inflation options. Indeed, they received the cashflows
of the linkers via swaps and for most linkers the final cashflows were
floored. This fixed not only the issue of hedging the pension fund
swaps but also, partly, the short YoY floor positions (Figure 8.14).
A long ZC inflation floor position partly hedges a short YoY floor
position. Intuitively, the CPI ratio of a ZC inflation swap is the product of CPI ratios of YoY Inflation swaps. Taking the logarithm of this
product, we get
ln
CPI(T )
CPI(0)
CPI(T )
CPI(T 1)
CPI(1)
= ln
+ ln
+ + ln
CPI(T 1)
CPI(T 2)
CPI(0)
161
i
i
i
perrucci 2012/7/11 17:12 page 162 #184
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 163 #185
ZC
YoY
Currency
Linear
Non-linear
Linear
Non-linear
Euro
Sterling
US dollar
Good
Good
Good
Medium
Low
Low/medium
Low
Low
Low
Good
Medium
Medium
i
i
i
perrucci 2012/7/11 17:12 page 164 #186
INFLATION-SENSITIVE ASSETS
Figure 8.17 BEI rates for 1Y, 5Y, 10Y and 30Y ZC swaps: (a) EMU
HICP and (b) US CPI
4.0
(a)
3.5
3.0
2.5
2.0
1.5
1.0
1Y
5Y
10Y
30Y
0.5
0
0.5
Nov Nov Nov Nov Nov Nov Nov Nov Nov
2003 2004 2005 2006 2007 2008 2009 2010 2011
4.5
(b)
2.5
0.5
1.5
3.5
5.5
Mar
2004
1Y
5Y
10Y
30Y
Mar
2005
Mar
2006
Mar
2007
Mar
2008
Mar
2009
Mar
2010
Mar
2011
i
i
i
perrucci 2012/7/11 17:12 page 165 #187
Figure 8.18 Realised volatility for forward BEI rates (1Y BEI, the 1Y in
2Y BEI and the 1Y in 10Y BEI) for (a) the EMU HICP and (b) US CPI
inflation swaps
4.0
3.5
10Y CPI
2Y CPI
1Y CPI
3.0
2.5
2.0
1.5
1.0
0.5
(a)
0
Nov
Nov
2003 2004
12
10
Nov
2005
Nov
2006
Nov
2007
Nov
2008
Nov
2009
Nov
2010
Nov
2011
10Y CPI
2Y CPI
1Y CPI
8
6
4
2
(b)
0
Mar
Mar
2004
2005
Mar
2006
Mar
2007
Mar
2008
Mar
2009
Mar
2010
Mar
2011
Looking at Figure 8.18 we can see that the peak of volatility occurs
in 2008.
Note that the YoY volatility presents some jumps; this is partly due
to the fact that the forward curve is extracted from the zero-coupon
inflation curve. In spite of the smoothing techniques used to derive
this curve, an important issue is the difference of liquidity between
the points of the curve. For example the 10Y zero-coupon swap rate
is certainly the most liquid point, but the 11Y rate is not so liquid
and its level is also dependent on the interpolation methods. These
165
i
i
i
perrucci 2012/7/11 17:12 page 166 #188
INFLATION-SENSITIVE ASSETS
(a)
0
10
15
20
25
30
3.1
2.6
2.1
%
1.6
1.0
(b)
0.6
10
15
20
25
30
i
i
i
perrucci 2012/7/11 17:12 page 167 #189
ln
CPI(t)
CPI(0)
= (rn rr )t
i
i
i
perrucci 2012/7/11 17:12 page 168 #190
INFLATION-SENSITIVE ASSETS
Real
economy
100
Spot FX rate
CPI(0)
Forward FX rate
CPI(t )
Nominal
economy
100er,t CPI(t )
100 CPI(0)
Non-arbitrage
100er,t CPI(0)
Fisher
equation
it is not surprising that the first models dealt with the inflation structure characteristics of TIPS. In their paper, Jarrow and Yildirim fitted their model to TIPS prices, nominal Treasuries prices and the
US CPI. Next, they tested the model for pricing TIPS via its hedging performance and, finally, they used it to price a ZC inflation call
option.
Their approach is an interesting one, as the methodology is based
on an analogy with a currency model. It is illustrated in Figure 8.20.
The CPI acts as an exchange rate between the nominal economy
and the real economy. Indeed, the forward CPI(t) is defined by nonarbitrage conditions. Specifically, in the real economy, $100, say, is
invested in a deposit at a real rate rr (dotted arrows) until maturity,
when the proceeds are converted to the nominal economy at a prefixed forward FX rate of CPI(t). Alternatively, $100 is converted into
the nominal economy at the spot FX rate CPI(0) (solid arrows), proceeds are invested at the nominal rate rn . In a non-arbitrage world,
the two investments must have the same return, leading us to define
the forward CPI ratio, which measures the return due to inflation as
the differential return between the nominal and real rates; this is
precisely the Fisher equation.
Jarrow and Yildirim added a stochastic component to both nominal and real rates, as well as the currency/inflation index. The
168
i
i
i
perrucci 2012/7/11 17:12 page 169 #191
i
i
i
perrucci 2012/7/11 17:12 page 170 #192
INFLATION-SENSITIVE ASSETS
uted;
the CPI acts as an exchange rate between the nominal and the
real economy.
We refer the reader to the appendix at the end of the chapter for
the pricing formulas of both nominal and real bonds, and also the
relationships linking the old and new set of variables. Note that Brigo
and Mercurio (2006) also introduced a model of both nominal and
real interest rates based on the currency analogy, which is similar to
this one.
Market models of inflation
The inflation derivatives market developed a couple of years after
the seminal model of Jarrow and Yildirim. However, the original
approach was adapted to the TIPS, but some issues emerged in using
this approach for some of these new derivatives instruments, such
as (see, for example, Belgrade et al 2004; Mercurio 2005)
non-observable parameters: inflation rates are derived from
real rates, but the latter are not directly observable from
inflation-linked bonds, for the reasons mentioned earlier,
there is no obvious link between ZCISs and YYISs,
the natural discount curve for the options and swaps is the
Libor curve.
As a result, new inflation models were devised that mirrored the
techniques used in Libor market models, and are based on information from the ZCISs and the YYISs in a consistent way. These models
focused on a set of forward inflation indexes for selected maturities
Ti , i = 1, . . . , N, ie, CPI(t, Ti ) as a lognormal processes under the
risk-neutral probability Q
dCPI(t, Ti )
Q
= (t, Ti ) dt + (t, Ti ) dWi ,
CPI(t, Ti )
i = 1, . . . , N
i
i
i
perrucci 2012/7/11 17:12 page 171 #193
The model allows for a much richer correlation structure among the
Q
Q
(N + 1) Brownian motions (Wi , i = 1, . . . , N; WB ), than what is
possible in the JarrowYildirim three-factor model.
ZC and YoY payouts can be written in terms of the future price
index levels or, alternatively, in terms of the forward index. For example, a zero-coupon T-maturity payout, be it a swap or an option, is
a function of the ratio
CPI(t, T )
CPI(T0 )
where the denominator has been fixed at trade inception, while the
numerator is unknown before T, and will be determined by the
stochastic evolution of the price index CPI(t). The latter can be written in terms of the stochastic evolution of the forward price index
CPI(t, T ) instead, as the two will coincide at t = T. Pricing instruments based on ZC payouts requires taking the risk-neutral expectations of the discounted terminal payouts. An opportune choice of
numeraire, ie, adopting the T-forward measure, further simplifies
things, as the forward price index CPI(t, T ) is, by construction, a
martingale under such a measure.
Things are a bit more complicated for YoY instruments, where the
terminal payout depends on the ratio of two price index levels, both
unknown at
CPI(t, T2 )
CPI
t < T1 < T2
CPI(T1 );
( T2 )
CPI(t, T1 )
Now there is no clever numeraire choice that can make this ratio a
martingale. In other words, convexity adjustments are unavoidable
when calculating the risk-neutral expectation of the discounted terminal payout. Specifically (neglecting lag effects), these will arise
from the covariance of the two forward price indexes as well as the
covariance of inflation with the nominal discounting entering the
expectation.
As a result, the discounted payment of the YoY(T, T + 1) =
CPI(T + 1)/CPI(T ) 1 must be adjusted for the covariance between
the inflation index at the two different times, as well as the covariance between the forward discount bond B(T, T + 1) and the forward
inflation index.
Mercurio (2005) adopted an alternative market model where both
real and nominal Libor forward rates for a set of maturities Ti , i =
1, . . . , N (thus, 2N Brownian motions) were modelled as lognormal
processes, and inflation derived from these.
171
i
i
i
perrucci 2012/7/11 17:12 page 172 #194
INFLATION-SENSITIVE ASSETS
Using these models, we can derive explicit formulas for inflation options that are simpler and more intuitive to calibrate than
those derived within currency-analogy-based models. Indeed, the
advantage of such models is to allow a rich dependence structure
between YoY instruments. This has become important as YoY and
ZC options evolve separately (different players and different flows),
which makes this dependence structure more and more complex. In
addition, it is usually simpler to carry out historical analysis on YoY
data, as there is a relatively long sample which can be used to develop
intuition about the volatility structure (Figures 8.17 and 8.18).
CONCLUSIONS
The inflation swaps market developed on the back of the hedging
activities of pension funds. Although pension funds initially bought
inflation-linked bonds, they readily switched to inflation swaps,
which could be more easily tailored to the specificity of their liabilities. Naturally, an interbank swap market soon developed, for which
the standard became the zero-coupon inflation swap, the structure
of choice of pension funds.
Another important step for the inflation market was the development of inflation-structured notes, which paid coupons linked to the
year-on-year inflation, and were popular among retail investors in
particular. These instruments were key in the development of YoY
inflation swaps and specifically YoY options, as well as an interbank
market for these instruments. In fact, the YoY format is the standard
format for inflation options, although the zero-coupon structure also
trades, with less liquidity. Using the example of inflation swaps, we
showed how it is not straightforward to hedge YoY swaps with ZC
swaps, as one has to take into consideration important convexity
adjustments.
In terms of options, one of the first models was the Jarrow
Yildirim model, which was calibrated using TIPS prices and historical inflation data. This is model was based on the currency analogy, with the inflation index modelled as an exchange rate between
nominal and real bonds. Soon, in part motivated by the YoY option
standard, new approaches came about, which modelled the forward
inflation index in its forward measure (similarly to Libor-market
models) and could be calibrated using inflation and nominal swaps
and YoY inflation options.
172
i
i
i
perrucci 2012/7/11 17:12 page 173 #195
Although this chapter has covered the key developments in inflation market in Europe and the US, inflation instruments (cash and
derivatives) have expanded in other countries as well. Inflationlinked bonds have been issued since the 1960s in some Latin American countries. There is an active inflation-linked bond and swap
market in Australia. Thailand and South Korea have both issued
inflation-linked bonds, and products will keep expanding as their
markets evolve and their pension systems reach financial maturity.
In any case, we expect the stepping stones to be the same: first, the
development of an inflation-linked bond market as the fundamental building block, followed by inflation derivatives, ie, swaps and
possibly options.
APPENDIX
JarrowYildirim currency-analogy-based model
Using the nominal zero-coupon bond as numeraire and its associated probability QN , the dynamics of the prices of the zero-coupon
nominal (N) and real (R) bonds are given by the following equations
(we use a time-decaying volatility model)
dBN (t, T )
N,QN
= rN (t) dt + N (t, T ) dWB (t)
(8.1)
BN (t, T )
dBR (t, T )
R,QN
= (rR (t) R,CPI (t)R (t, T )) dt R (t, T ) dWB (t)
BR (t, T )
(8.2)
x (t, T ) = x (t)
1 exp(x (T t))
(8.3)
The CPI acts as an exchange rate between the nominal and the real
economy
dCPI(t)
CPI,QN
= (rN (t) rR (t)) dt + CPI dWt
CPI(t)
CPI,QN
dWt
N,QN
dWt
CPI,QN
= N,CPI dWt
R,QR
dWt
(8.4)
(8.5)
i
i
i
perrucci 2012/7/11 17:12 page 174 #196
INFLATION-SENSITIVE ASSETS
1 ex
x
(8.6)
(8.7)
(8.8)
(8.9)
The change from the old projection into the new projection is based
on the following equations
i = N2 + R2 2N,R N R
N
R
dWti =
dWtN
dWtR
i
i
(8.10)
(8.11)
In a defined benefits pension fund, the benefits payable can be a function of several factors,
such as earnings history, years worked and age, but do not depend on the funds investment
returns.
The Japanese case is an exception, as Japan experienced a long deflation period. Note
that Japanese government inflation-linked bonds do not have an embedded inflation floor,
although some banks have structured inflation-linked notes with such a floor.
REFERENCES
Belgrade, N., E. Benhamou and E. Koehler, 2004, A Market Model for Inflation, URL:
ftp://mse.univ-paris1.fr/pub/mse/cahiers2004/B04050.pdf.
Belgrade, N., E. Benhamou and E. Koehler, 2005, Modelling Inflation in Finance,
Inflation-linked Products, Chapter 6 (London: Risk Books).
Bnaben, B. (ed), 2005, Inflation-linked Products (London: Risk Books).
Bnaben, B., and H. Cros, 2008, Global Inflation Derivatives Markets, in Inflation Risks
and Products, Chapter 11 (London: Risk Books).
Bnaben, B., and S. Goldenberg (eds), 2008, Inflation Risk and Products (London: Risk
Books).
Bnaben, B., and N. Tabardel, 2008, Inflation-Linked Options, in Inflation Risk and
Products, Chapter 15 (London: Risk Books).
Brigo, D., and F. Mercurio, 2006, Interest Rate Models: Theory and Practice with Smile, Inflation
and Credit, Part VI, Second Edition (Springer).
Davis, E. P., 2000, Regulation of Private Pensions: A Case Study of the UK, Discussion
Paper PI-0009, The Pensions Institute, URL: http://www.ephilipdavis.com/wp0009.pdf.
174
i
i
i
perrucci 2012/7/11 17:12 page 175 #197
European Central Bank, 2000, Seasonal Adjustment of Monetary Aggregates and HICP
for the Euro Area, Aggregates and HICP for the Euro Area, August (Extract).
Jarrow, R., and Y. Yildirim, 2003, Pricing Treasury Inflation Protected Securities and
Related Derivatives using an HJM model, Journal of Financial and Quantitative Analysis
38(2), pp. 337358.
Mercurio, F., 2005, Pricing Inflation-Indexed Derivatives, Quantitative Finance 5, pp. 289
302.
Mercurio, F., and N. Moreni, 2006, Inflation with a Smile, Risk, March, p. 70.
Peng, W., 2006, Understanding Inflation Convexity, IXIS Capital Markets, URL: http://
inflationinfo.com/CPICvx.pdf.
Taylor, J. B., 1993, Discretion versus Policy Rules in Practice, Carnegie-Rochester Conference
Series on Public Policy, Volume 39, pp. 195214 (Elsevier).
Wilson Committee, 1980, Report of the Committee to Review the Functioning of Financial
Institutions (Chairman Sir Harold Wilson), Command Paper 7937 (London: HMSO).
175
i
i
i
perrucci 2012/7/11 17:12 page 176 #198
i
i
i
perrucci 2012/7/11 17:12 page 177 #199
Part II
i
i
i
perrucci 2012/7/11 17:12 page 178 #200
i
i
i
perrucci 2012/7/11 17:12 page 179 #201
i
i
i
perrucci 2012/7/11 17:12 page 180 #202
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 181 #203
Figure 9.1 Reserve Bank of New Zealand 90-day interest rate forecast
Projection
10
9
8
7
6
June MPS
5
4
Central
3
2
1
0
2004
2006
2008
2010
2012
shock occurs. Third, IT is more transparent and consistent, as monetary policy is committed to targeting a specific price level or range
of price levels. Typically, the specific target is set in terms of the
headline Consumer Price Index (CPI) or core CPI, or the headline
or core Personal Consumption Expenditure (PCE) prices index. The
target itself is most often set not by the central bank alone, but rather
by the government or by agreement by both. The central bank then
commits to using available policy instruments (almost exclusively
the short-term interest rate, such as the two-week repo) to respond
to economic shocks and induce inflation to return to target over a
medium run.3
The success of IT hinges on the capacity of the central bank to react
with the appropriate policy instrument to developments expected
in the future. Todays actions have to ensure that inflation returns to
the target level at a given horizon, given the most plausible assumptions. In doing so, small short-term disturbances are often ignored,
as the focus is on the medium-to-long-term risk of inflation. Obviously, this requires a good understanding of policy transmission and
its lags (which typically range between four and eight quarters, ie,
one to two years). Such a task puts enormous pressure on forecasting capacity, and on the ability to make decisions under uncertainty.
Almost any central bank produces various kinds of economic forecasts, but many use them only as an indicative macroeconomic outlook and in conjunction with several other inputs. By contrast, in an
181
i
i
i
perrucci 2012/7/11 17:12 page 182 #204
INFLATION-SENSITIVE ASSETS
75%
50%
6
5
4
3
2
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
Figure 9.3 Norges Bank policy rate forecast and confidence bands (in
percent)
9
8
30%
50%
70%
90%
7
6
5
4
3
2
1
0
2008
2009
2010
2011
2012
2013
2014
IT regime, forecasts become an essential backbone for the decisionmaking process. Furthermore, there is an intricate complexity about
these forecasts. Their main focus is typically not directly on inflation but on the trajectory of interest rates. If you look at the inflation
forecast charts that IT central banks produce every quarter, you will
note that it is not a very interesting chart: inflation always returns to
the target level sooner or later. The more important chart shows the
trajectory of interest rates that is consistent with this inflation profile,
ie, the trajectory consistent with bringing inflation to target under
182
i
i
i
perrucci 2012/7/11 17:12 page 183 #205
Figure 9.4 Czech National Bank three-month Pribor rate forecast and
confidence bands (in percent)
5
90%
70%
50%
30%
4
3
2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
Figure 9.5 Bank of Israel interest rate forecast and confidence bands
(percentage)
7
6
5
4
3
2
1
0
2010
2011
2012
i
i
i
perrucci 2012/7/11 17:12 page 184 #206
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 185 #207
why the bank is doing it and how the bank learns from its own
mistakes (accountability), it is much easier to affect and ultimately
anchor expectations on the inflation target. In turn, if expectations
are anchored, then the actual outcomes will become more favourable
and stable.
This new approach to policy making could not be in a starker
contrast with the older paradigm. In the past, monetary policy was
believed to be most efficient if it was based on unexpected surprises
(in fact, there is still a relatively large amount of academic research
nowadays that finds the so-called money supply shocks the proper
monetary policy tool to investigate), and assumptions about systematic behaviour of policy rates had no role in economic projections.
There also seems to be a widening gap between the practices of
banks that have an explicit inflation target and those who do not,
most notably the central banks behind the two major currencies: the
European Central Bank and the US Federal Reserve System. It is hard
to believe that the first ever press conference by the US Feds Chairman occurred only in April 2011, and documents explaining the Fed
staff projections were only released with a six-year lag, until the decision to make forecasts available to the public in January 2012. Until
then, the US Fed communication strategy had been in sharp contrast
with one of the other central banks, for which making regular public
appearances, publishing full-fledged reports immediately after policymaking meetings and posting models and other analytical tools
on websites had been routine practice for a long time. There can be
no wonder that markets were very surprised and rather confused to
hear Ben Bernanke say in his press release4 that
economic conditions are likely to warrant exceptionally low
levels for the federal funds rate for an extended period.
i
i
i
perrucci 2012/7/11 17:12 page 186 #208
INFLATION-SENSITIVE ASSETS
Issing 2008). Instead, the bank relies officially upon a so-called twopillar strategy, consisting of economic analysis and monetary
analysis. This gives its policy a traditional money-targeting flavour
to some extent. The ECB staff finds, nevertheless,
the assumptions about short-term interest rates [to be] of a purely
technical nature.
(European Central Bank 2011)
i
i
i
perrucci 2012/7/11 17:12 page 187 #209
Reliability
(prediction power)
Short term
Medium term
Projection horizon
i
i
i
perrucci 2012/7/11 17:12 page 188 #210
INFLATION-SENSITIVE ASSETS
Initial conditions
External
projections
Nowcasts and
near-term
forecasts
Trend-cycle
analysis
Trend
assumptions
Core
projection
model
Satellite
models
Informal
judgement
Judgemental adjustments
i
i
i
perrucci 2012/7/11 17:12 page 189 #211
and policymakers. Given that the model is used to guide the policymaking process itself, this would seem to imply that monetary policy
and expectations are exogenous to it. Yet, a considerable number of
central banks, including some in advanced economies, still struggle
to understand how crucial an assumption this is for good conduct
of their policies. These issues will be discussed in greater detail in
the next section.
Models for the so-called now-casts deliver estimates for the current month or current quarter of those variables that are not yet
available due to a publication lag. Techniques used in now-casting
are basically the same as in near-term forecasting, which produces
a very short-term outlook for the main variables, say a quarter or
two quarters ahead. These numbers are then commonly treated as
hard data when entered into the core projection model (effectively
extending the actual time-series observations). Most of the central
banks combine two basic approaches to constructing now-casts and
near-term forecasts (NTFs): expert knowledge-based projections and
reduced-form multivariate time-series models. In both, the estimates
and forecasts are built bottom-up using an extensive number
of detailed macroeconomic indicators (most of which do not even
appear in the core projection models).8 The multivariate reducedform models include various sorts of estimated vector autoregressions (VARs) adjusted to deal with the problem of potential overfitting, such as Bayesian vector autoregressions (BVARs) or factoraugmented vector autoregressions (FAVARs); see the appendix on
page 201 for a brief explanation of these two techniques.
Trend-cycle analysis9 is needed because the core projection models are frequently built around economic concepts (such as equilibrium values) that are not directly observable: two examples are the
output gap (or, equivalently, potential output) and the natural rate
of interest. Trend-cycle analysis is therefore necessary for setting up
the initial conditions for the model. Broadly speaking, there are three
options available.
1. Univariate filters, which produce more or less mechanical
results with the option of judgemental adjustments. These
include the popular Leser (1961) filter, also known as the
Hodrick and Prescott (1997) filter in economics.
2. Various multivariate filters relating the trends and cycles in the
variable(s) in question to other indicators that are expected to
189
i
i
i
perrucci 2012/7/11 17:12 page 190 #212
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 191 #213
rative;
policy recommendations consistent with, and incorporated in,
i
i
i
perrucci 2012/7/11 17:12 page 192 #214
INFLATION-SENSITIVE ASSETS
making clear who in fact owns the forecast (in other words, whether
it expresses staff views or policymakers views) and its weight in
the decision-making. Obviously, the ownership of the forecast also
influences the forecasting process itself. If it is to be a policymakers
forecast, the production will almost surely take longer, as the central bank staff needs to meet with the policymakers more frequently,
and eventually fine-tune the results to their liking. Indeed, at some
central banks, there are two distinct forecasting rounds: the first produces projections based on staff views, while the second incorporates
policymakers assumptions and judgement. When it comes to the
issue of forecast ownership and the weight that forecasting has on
policy, different central banks take different approaches, which are
clear for some and more ambiguous for others. A few examples are
listed below.
The Bank of Canadas Monetary Policy Report, July 2011,
i
i
i
perrucci 2012/7/11 17:12 page 193 #215
i
i
i
perrucci 2012/7/11 17:12 page 194 #216
INFLATION-SENSITIVE ASSETS
Long-term neutrality. Real outcomes (as opposed to monetary, or nominal, variables) remain unaffected by monetary policies in the long run (unless the policies are very
bad) and, vice versa, the central bank can set and achieve
any inflation target (or other nominal, or monetary, target)
without distorting real outcomes. In practice, of course,
the neutrality is a convenient assumption only within a
range of relatively low and stable inflation rates.
Medium-term real effects of monetary policy. Unlike in
the long run, the central banks actions do impact on real
outcomes in the medium run, thanks to various types
of frictions and rigidities (such as price and wage stickiness). These rigidities wear off after a sufficiently long
time, allowing neutrality to reign again.
The stabilising role of monetary policy throughout the
business cycle. The systematic reaction of the central bank
to macroeconomic shocks must be sufficient to prevent an
inflation (or deflation) spiral. For example, an inflation
spiral occurs whenever high inflation, not fought against
by the central bank, feeds into expectations, thus reducing
real rates and raising demand. Excess demand adds more
pressure still on inflation, and the economy is caught in a
vicious circle.
Forward-looking expectations. Although it would be
unrealistic to assume fully forward-looking expectations
in practical applications, the rational element in expectations must not be neglected. Otherwise, the model-based
advice could easily make the policymakers believe they
could exploit permanent trade-offs where there are none;
in other words, the model would imply the central bank
could fool all of the people all of the time.
Robustness to policy errors: the predictive power of core pro-
jection models is often defined by their ability to get the timing of the turning points in a business cycle right. This proves
more important than any other statistical test. Accurate turning points are one of the preconditions for avoiding systematic policy mistakes: such as type A policy errors (being too
ahead of the curve, ie, reacting to false signals) or type B
194
i
i
i
perrucci 2012/7/11 17:12 page 195 #217
policy errors (being behind the curve, ie, not reacting to true
signals).
Different types of core projection models
We can identify four broad varieties of core projection models:
1. large-scale econometric models;
2. weak-form dynamic stochastic general equilibrium (DSGE)
models;
3. semi-structural flow (or gap) models; and
4. semi-structural stock-flow models.
Large-scale econometric models are considered to be outdated in
the context of monetary policy models (which is not to say they are
useless, but that they are not the right model for monetary policy).
They are usually designed bottom-up, with great emphasis on disaggregation; they also most often lack an explicit role for monetary
policy or expectations, with the notable exception of the US Federal
Reserve staffs model.13 Because they are estimated without proper
monetary policy, they tend to incorrectly interpret various monetary
stabilities or instabilities observed in the data.
The other three types of model are generally much more compatible with modern monetary policymaking. There exists, though,
a certain trade-off between operational simplicity and theoretical
coherence across those three varieties. We now briefly explain some
of their defining features.
DSGE models start by making explicit and detailed assumptions
about what types of economic agent act in the model economy (such
as households, producers, retailers, banks, the central bank and government), what their objectives are (to maximise utility or profits)
and the environment in which they interact (eg, competitive markets,
monopolistic markets). Then, they derive the optimality conditions
for the behaviour of each, aggregate these over the entire economy
and impose general equilibrium conditions (ie, that demand equals
supply in each market at the prevailing price). However appealing the concept of a DSGE is, it is in practice impossible to build
DSGEs that perform well in all the desired empirical dimensions
while keeping all the underlying assumptions about the agents
objectives strictly consistent with microeconomic and behavioural
theories. Therefore, various shortcuts, simplifications and ad hoc
195
i
i
i
perrucci 2012/7/11 17:12 page 196 #218
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 197 #219
i
i
i
perrucci 2012/7/11 17:12 page 198 #220
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 199 #221
Supply shock
Imported inflation
Inflation expectations
Demand and others
Initial conditions
10
5
0
5
10
20
2005 Q1
2005 Q2
2005 Q3
2005 Q4
2006 Q1
2006 Q2
2006 Q3
2006 Q4
2007 Q1
2007 Q2
2007 Q3
2007 Q4
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
15
Source: OGResearch.
i
i
i
perrucci 2012/7/11 17:12 page 200 #222
INFLATION-SENSITIVE ASSETS
0.5
0
0.5
0.10
0.15
0.25
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
2016 Q3
2016 Q4
0.20
Source: OGResearch.
i
i
i
perrucci 2012/7/11 17:12 page 201 #223
Figure 9.10 Change in (a) baseline inflation and (b) real GDP forecast
due to a food price shock
(a)
14
Baseline
12
Scenario
10
8
6
4
2
0
2007 2008 2009 2010 2011 2012 2013 2014 2015
(b)
20
15
10
5
0
5
Baseline
10
Scenario
15
20
25
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
Source: OGResearch.
always be up to the analysts judgement to create the relevant scenarios and, in conjunction with all the available quantitative and
qualitative information, determine the best policy action path going
forward.
APPENDIX: MULTIVARIATE MODELS FOR NEAR-TERM
FORECASTING
In this appendix, we outline the methodology of Bayesian vector
autoregressions and factor-augmented vector autoregressions. Both
of these methodologies were developed specifically to deal with
empirical short-term dynamic relationships between a (potentially
very) large number of variables. Empirical models with a very large
201
i
i
i
perrucci 2012/7/11 17:12 page 202 #224
INFLATION-SENSITIVE ASSETS
(9.1)
ft = A1 ft1 + + Ak ftk + t
(9.2)
i
i
i
perrucci 2012/7/11 17:12 page 203 #225
inflation and foreign interest rates) and also stylised factors affecting
the cost of borrowing.
Glossary
Real gaps (percentage deviations in real variables from their
long-run trends):
yt
rt
zt
yt
qt
t
ut
output gap;
real interest gap;
real exchange gap;
foreign demand gap;
real price of capital gap;
external finance premium;
foreign exchange risk spread.
Nominal variables:
it
pt
t
t4
tm
it
p
t
Structural shocks:
y
t
t 1
t 2
ts
ti
q
t
The superscript e denotes expectations in general; the operator Et [] denotes the model-consistent (rational) expecta-
i
i
i
perrucci 2012/7/11 17:12 page 204 #226
INFLATION-SENSITIVE ASSETS
Identities between log price levels and inflation rates are not
listed here.
Real trends and the foreign exchange risk spread are exoge-
(9.3)
t = t 1 + (t 2 t21 )
(9.5)
(9.6)
i
i
i
perrucci 2012/7/11 17:12 page 205 #227
qt = rt + yte+1 + (1 )qet+1 + t
t = qt + t
(9.7)
(9.8)
zt = st + p
t pt z
rt = it te+1 rt
(9.9)
(9.10)
ta+1 = ta + (t ta )
(9.11)
Iceland: although at the time of writing in 2012 the inflation target is still technically in force,
the central bank is concentrating mostly on keeping the exchange rate stable.
Meyer (2001) has a nice discussion of central bank mandates and objectives from the perspective of IT.
Although the financial crisis revived the use of other instruments, such as FX interventions or
quantity based operations, these were employed either to support the transmission of interest
rate policy or to satisfy different objectives outside the IT framework, such as pro-exporting
industrial policy.
205
i
i
i
perrucci 2012/7/11 17:12 page 206 #228
INFLATION-SENSITIVE ASSETS
Such as an inflation forecast for a particular rather irrelevant sector, calculated and discussed
up to three decimal places.
In simple terms, reduced-form models explain a set of dependent variables in terms of a set
of independent variables (for example, in a regression framework). Structural models deal
with equilibrium relationships between supply and demand, which are likely to dominate in
the medium to long term.
For example, switching expenditures between cheaper and costlier goods, or deciding on
investment in productive capacity based on the cost of bank lending.
For instance, while the core projection model may work with the headline CPI and the five traditional demand components of GDP only (consumption, investment, government, exports,
imports), the now-casts and NTFs make use of a detailed sector-by-sector breakdown of the
CPI into sub-indexes, and both demand a production view of GDP flows.
Models for trend-cycle analysis are called filters because they aim at filtering out highfrequency fluctuations while identifying lower frequency relationships.
10 Non-linearities that may matter for monetary policy include an asymmetric, convex Phillips
curve (prices fall less in recessions than they rise in booms) and endogenous credibility (expectations are more difficult to anchor if the central bank has a poor track record in keeping
inflation low and stable).
11 Economic modelling software packages that offer a high degree of flexibility in incorporating
conditioning information and judgemental adjustments in simulations include the IRIS Toolbox (http://www.iris-toolbox.com), Troll (http://www.intex.com/troll) and Sirius (http://
www.ogresearch.com/products).
12 See http://www.federalreserve.gov.
13 The FRB/US model; see http://www.federalreserve.gov.
14 BEQM stands for Bank of England Quarterly Model. ToTEM stands for Terms of Trade Economic Model. KITT stands for Kiwi Inflation Targeting Technology. NEMO stands for Norwegian Economy Model. RAMSES stands for Riksbank Aggregate Macro-model for Studies
of the Economy of Sweden.
15 For instance, data on physical capital is not readily available or reliable even in many advanced
countries; the relationship between the countrys net investment position and the reported
current accounts tends to be fuzzy because of occasionally unclear definitions or hard-to-track
valuation effects.
REFERENCES
Banbura, M., D. Giannone and L. Reichlin, 2008, Large Bayesian VARs, ECB Working
Paper 966.
Benes, J., K. Clinton, R. Garcia-Saltos, M. Johnson, D. Laxton, P. Manchev and T.
Matheson, 2009, Estimating Potential Output with a Multivariate Filter, IMF Working
Paper 10/285.
Bernanke, B., J. Boivin and P. Eliasz, 2005, Measuring Monetary Policy: A FactorAugmented Vector Autoregression (FAVAR) Approach, Quarterly Journal of Economics
120(1), pp. 387422.
Bernanke, B., M. Gertler and S. Gilchrist, 1999, The Financial Accelerator in a Quantitative Business Cycle Framework, in J. B. Taylor and M. Woodford (eds), The Handbook of
Macroeconomics, Volume 1C (Elsevier).
Carabiencov, I., I. Ermolaev, C. Freedman, M. Juillard, O. Kamenik, D. Korshunov and
D. Laxton, 2008, A Small Quarterly Projection Model of the US Economy, IMF Working
Paper 08/278.
206
i
i
i
perrucci 2012/7/11 17:12 page 207 #229
Chari, V. V., 2010, Testimony before the Committee on Science and Technology, Subcommittee on Investigations and Oversight, US House of Representatives, June.
Chen, H., K. Clinton, M. Johnson, O. Kamenik and D. Laxton, 2009, Constructing
Forecast Confidence Bands during the Financial Crisis, IMF Working Paper 09/214.
European Central Bank, 2011, ECB Staff Macroeconomic Projections for the Euro Area,
September, URL: http://ecb.europa.eu/pub.
Faust, J., 2009, The New Macro Models: Washing Our Hands and Watching for Icebergs,
Sveriges Riksbank Economic Review, 2009(1).
Hodrick, R., and E. C. Prescott, 1997, Postwar US Business Cycles: An Empirical
Investigation, Journal of Money, Credit, and Banking 29, pp. 116.
Issing, O., 2008, In Search of Monetary Stability: The Evolution of Monetary Policy,
Contribution to Seventh BIS Annual Conference, June.
Leser, C. E. V., 1961, A Simple Method of Trend Construction, Journal of the Royal
Statistical Society, Series B 23, pp. 91107.
Meyer, L. H., 2001, Inflation Targets and Inflation Targeting, Speech at the University
of California at San Diego Economics Roundtable, July.
Norges Bank, 2011, Criteria for an Appropriate Interest Rate Path, Monetary Policy
Report 2/2011.
Solow, R., 2010, Building a Science of Economics for the Real World, Testimony before
US Congress Committee on Science and Technology, July.
Taylor, J. B., 1993, Discretion versus Policy Rules in Practice, Carnegie-Rochester Conference
Series on Public Policy, Volume 39, pp. 195214 (Elsevier).
Taylor, J. B., 1998, The Robustness and Efficiency of Monetary Policy Rules as Guidelines
for Interest Rate Setting by the European Central Bank, Seminar Paper 649, Institute for
International Economic Studies, Stockholm University.
207
i
i
i
perrucci 2012/7/11 17:12 page 208 #230
i
i
i
perrucci 2012/7/11 17:12 page 209 #231
10
i
i
i
perrucci 2012/7/11 17:12 page 210 #232
INFLATION-SENSITIVE ASSETS
strategies that start with zero value, have zero probability of loss
and have a positive probability of a positive profit at a later point in
time. Under the assumption that a market does not allow arbitrage
opportunities, default-free zero-coupon bond prices can be obtained
using the risk-neutral valuation approach and expressed as an expectation under the risk-neutral probability measure Q. If the market
is complete, ie, any derivative can be hedged using a self-financing
portfolio of market securities, then the above probability measure Q
not only exists but it is unique. These assumptions, together, result in
the well-defined, unique bond prices (see the appendix on page 235).
Namely, let Pt, denote the price at time t of the zero-coupon bond
maturing at time t + and let yt, be the corresponding yield. We
have
Q
Pt, = exp[yt, ] = Et
exp
t+
rs ds
The result above is quite elegant, and offers a potentially useful link
between the short rate r (which is influenced by monetary policy)
and the rest of the yield curve. However, the problem is that our
observations of the short rate r belong to the real world under the
(data-generating) probability measure (or physical measure) P. In
other words, by collecting data on the short rate (using for example
the one-month interest rate as a proxy) we might gain insight on
its dynamics under P and yet, to calculate bond prices, we have to
take expectations under Q. We need a way to relate expectations
under different (yet equivalent) measures to be able to bridge the
gap between the two worlds.
In between two worlds: the price of risk
As mentioned earlier, we need to have a model that captures the
empirical properties of the data under the physical measure P. The
no-arbitrage assumption, which ensures the existence of Q also
ensures the existence of a transformation between Q and the real
data-generating measure P (two equivalent probability measures).1
Specifically
Q
t Et [ ] = EPt [T ]
where t is the stochastic process, representing the change in probability density between the two measures. Specific functional forms
are assumed for t , both to ensure that, being a probability density transformation, it is positively defined, and in order to preserve
210
i
i
i
perrucci 2012/7/11 17:12 page 211 #233
Data-generating
measure
[ ]
Price of
risk
EQt []
EtP
Pt, = Et
= EPt
t+
exp
rs ds
t+
T
exp
rs ds
t
t
i
i
i
perrucci 2012/7/11 17:12 page 212 #234
INFLATION-SENSITIVE ASSETS
the most salient empirical features of the US term structure are that
the nominal yield curve is on average upward-sloping,
the standard deviation of yields decreases with bond maturity,
yields are highly autocorrelated,
term premiums change over time,
yields are not normally distributed.
EPt
t+
t
rs ds
212
i
i
i
perrucci 2012/7/11 17:12 page 213 #235
Clearly, this does not hold in general except when interest rates are
certain, as
there is in general a non-zero price of risk, s 0, and
even if the price of risk were zero, so that investors are risk-
t+
EPt exp
rs ds
def
EPt
t+
t
+ Jensens terms
rs ds
drt = ( rt ) dt + dBt
Q
drt = ( rt ) dt + rt dBt
i
i
i
perrucci 2012/7/11 17:12 page 214 #236
INFLATION-SENSITIVE ASSETS
Instantaneous
short rate
Xt = (X1,t ,,XN,t)
rt = R(Xt ,t)
dBPt + t dt = dBt
i
i
i
perrucci 2012/7/11 17:12 page 215 #237
under P
Q
are all assumed to be affine functions of the state variables (the superscript T indicates transposition). In such models, bond yields also
215
i
i
i
perrucci 2012/7/11 17:12 page 216 #238
INFLATION-SENSITIVE ASSETS
In this case, the state vector dynamics is clearly affine under both Q
and P measures. See the appendix on page 240 for explicit solutions
for yields and bond price in this model.
However, note that for an N-dimensional state vector, the equations above introduce a large number of parameters to be estimated
(Table 10.1). This poses a common empirical challenge for all multifactor term structure models. For example, for the case of N = 3,
there are a total of 37 parameters to be determined.
Explicit solutions for yields and bond price for this simple model
are derived in the appendix on page 240.
Stochastic volatility models
The volatility matrix driving the evolution of the state vector does
not need to be constant as in the simple AR(1) model introduced
above, and can, in general, be a function of time and the state vector
itself. However, in affine term structure models, the variance needs
216
i
i
i
perrucci 2012/7/11 17:12 page 217 #239
0
X
0
X
N
N2
N2
1
N
N
N2
Total
3N 2 + 3N + 1
s(Xt ) = 0 s0 + sTx Xt
s
+ sT1x Xt 0
0
10
..
.
0
0
0
s(Xt ) =
..
..
.
.
0
0
sN0 + sTNx Xt
where s10 , . . . , sN0 are scalars and sT1x , . . . , sTNx are 1 N row vectors.
As mentioned earlier, to ensure that the dynamics is affine under
both Q and P measures, and that bond yields are affine, the product
(Xt , t) t should be affine (see also the appendix on page 240 on
this point). This implies that in stochastic volatility term structure
models, the market price of risk is not affine. The most common
functional assumption for the market price of risk, albeit not the
most general one (Dai and Singleton 2000), is
t = s(Xt )0
i
i
i
perrucci 2012/7/11 17:12 page 218 #240
INFLATION-SENSITIVE ASSETS
where
i
i
i
perrucci 2012/7/11 17:12 page 219 #241
i
i
i
perrucci 2012/7/11 17:12 page 220 #242
INFLATION-SENSITIVE ASSETS
time window used for estimation. Bond data used to estimate the
five-factor term structure model are monthly nominal zero-coupon
yields (continuously compounded) over the period 19522000. By
using the variance decomposition, Ang and Piazzesi show that the
two macro factors are major drivers of yields, especially in the short
end of the yield curve, while the latent factors tend to dominate in
the long end of the curve.
To summarise, in addition to the aforementioned studies, the
1990s and early 2000s produced a wealth of research regarding the
information contained in the term structure of nominal interest rates.
The interested reader can refer to a thorough treatment of DTSMs
and their estimation by Singleton (2006) and a comprehensive review
by Piazzesi (2010) for a more complete reference.
THE US TIPS MARKET
To a large extent, interest in the modelling of real interest rates has
been fostered by the development of the market for US Treasury
Inflation-Protected Securities (TIPS), launched in 1997. Since then,
government bond markets, the market for US Treasury nominal debt
and the market for inflation-indexed debt have provided an excellent laboratory for studying macroeconomic issues such as inferring inflation expectations, estimating inflation risk premiums and
extracting the probability of deflation. See Campbell et al (2009) for a
detailed and comprehensive overview of inflation-indexed markets
in both the US and the UK.
Initially the TIPS market did not attract the attention of many
researchers, partly due to its liquidity problems, and partly because
inflation (or disinflation) concerns were not as common in late 1990s
as they are at the time of writing. However, in the late 200s this
changed, in part because rising global risks contributed to a flightto-quality from riskier equities markets to safer markets such as
the markets for US Treasury nominal and inflation-indexed debt.
For instance, according to Morningstar (2010), total net asset values
of TIPS funds increased by more than 54% (or about US$19.5 billion)
over the one-year period from January 2009 to January 2010.
The most important feature of TIPS is that their principal and, consequently, the coupon payments are linked to the value of the Consumer Price Index (CPI). As such, these payments are denominated
in real rather than nominal terms, and thus TIPS can be considered
220
i
i
i
perrucci 2012/7/11 17:12 page 221 #243
i
i
i
perrucci 2012/7/11 17:12 page 222 #244
INFLATION-SENSITIVE ASSETS
Figure 10.3 Ten zero-coupon nominal and TIPS yields and inflation
break-even rates
8
6
%
4
(a)
2
2000
2002
2004
2006
2008
2010
2012
2002
2004
2006
2008
2010
2012
2002
2004
2006
2008
2010
2012
6
4
%
2
(b)
0
2000
3
2
%
1
(c)
0
2000
(a) Ten-year nominal yields; (b) ten-year TIPS yields; (c) ten-year inflation compensation.
Source: Grishchenko and Huang (2010).
EPt (It,R ) +
where Tt,R denotes the real interest rate risk premium and IRPt, is
the inflation risk premium.
As mentioned earlier, the break-even inflation rate, BEIt, , ie, the
difference between nominal and TIPS yields, represents the amount
of inflation compensation that investors in the nominal Treasury
222
i
i
i
perrucci 2012/7/11 17:12 page 223 #245
EPt [It,inf
] + IRPt,
i
i
i
perrucci 2012/7/11 17:12 page 224 #246
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 225 #247
nominal rates. The third latent factor is interpreted as a level factor for survey inflation expectations (survey-based inflation expectations are formed in the real-world probability). While inflation
affects the nominal curve, this latent survey factor does not (it is
hidden), and it represents information about the expected path of
inflation, which is not captured by inflation itself. In addition to quarterly GDP and CPI data, both nominal and real (ie, US TIPS) yields
are used in the calibration, as well as inflation expectations from
three popular market surveys (the Livingston Survey, the Survey
of Professional Forecasters and the Blue Chip Economic Indicators
Survey)10 . The survey factor and survey inflation data are found
to greatly increase out-of-sample forecasting of future inflation. The
study also points out the following two consequences of using US
TIPS data (from 2003 onwards):
1. the average level of real yields is higher (implying a lower
average inflation risk premium);
2. real rates are more volatile (implying a decline in the volatility
of the inflation risk premium).
The above findings might partly be caused by liquidity problems
on the TIPS market. For example, higher real yields might include
a liquidity premium that would result in the downward bias of
the average inflation risk premium. Therefore, liquidity should be
accounted for when TIPS yields are modelled. DAmico et al (2009)
were among the first to take into account TIPS liquidity in a model of
both nominal yields and TIPS yields. They found that, while three
principal components explain over 97% of weekly nominal yield
changes, a fourth factor is needed to explain the changes of both
nominal yields and TIPS yields. The latter factor can be interpreted
as a liquidity premium in TIPS yields, while the remaining three are
the usual level, slope and curvature (latent) factors driving both yield
curves. Thus, they use a four-factor Gaussian term structure, where
nominal and US TIPS yields are modelled jointly in order to estimate
the TIPS liquidity premium, expected inflation and inflation risk premium. DAmico et al show that ignoring the liquidity component can
lead to severely biased estimates of expected inflation and inflation
risk premiums. Specifically, if TIPS yields are identified with real
rates, model-implied inflation expectations do not fit the empirical
downwards trend observed in inflation survey data over time, and
225
i
i
i
perrucci 2012/7/11 17:12 page 226 #248
INFLATION-SENSITIVE ASSETS
the estimates of the inflation risk premium are low or even negative.
When the liquidity factor is introduced, the DAmico et al model
is able to reproduce the downwards trend in expected inflation, as
well as positive inflation risk premium. Consequently, a change in
TIPS break-even rates cannot simply be interpreted as a change in
inflation expectations, as both the inflation and the liquidity risk
premiums play important roles.
Adrian and Wu (2010) used a five-factor model with two inflation
factors, two real factors and a factor that governs the dynamics of
variances and co-variances of state variables. They found that inflation expectations differ significantly from the break-even inflation
rate when inflation volatility is high. Chen et al (2010) estimated a
two-factor term structure model with real rates and expected inflation as state variables, and found that the expected inflation is flat,
while the inflation risk premium is upward sloping (in line with Ang
et al (2008)).
Haubrich et al (2011) estimated a seven-factor term structure
model, using nominal Treasury yields, inflation survey forecasts
(from the Blue Chip Economic Indicators Survey and the Survey of
Professional Forecasters), realised inflation rates and zero-coupon
inflation swaps rates. Their data span the period from January 1982
to May 2010 (although the inflation swap data only starts in April
2003). Inflation swaps are the most liquid inflation derivative contracts and are quoted for maturities ranging from one year to 30 years
(see Chapters 7 and 8 herein). They assume that nominal and real
yields are driven by three state variables that represent the shortterm real interest rate, expected inflation and long-run inflation
(what they call inflations central tendency), in addition to four
volatility (price of risk) factors, which are a mix of normal and chisquared innovations, follow Garch processes and determine bond
risk premiums. Haubrich et al found that the short real interest rate
is the most volatile component of the yield curve and displays significant mean reversion. Expected inflation over short horizons was
also found to be volatile, negatively correlated to the real rate and
mean reverting. Long-term inflation expectations declined substantially over the 19822010 sample period, consistent with the Federal
Reserves credibility in maintaining price stability. The study also
found evidence of a real interest rate term premium Tt,R , which was
substantial, averaging 102 basis points (bp) and varying between
226
i
i
i
perrucci 2012/7/11 17:12 page 227 #249
87 and 121bp for the 10-year zero-coupon bonds over the sample
period. The 10-year inflation risk premium has an upward-sloping
term structure, with the average of 42bp and varying between 23
and 55bp over the sample period. Incidentally, the real interest rates
risk premium was also found to be predominant in the eurozone
area (Hrdahl and Tristani 2007), although an inflation premium
was also present.
According to Haubrich et al, one advantage of their approach is
that the inflation expectations and real yields obtained from their
model are more accurate, because inflation swap rates are less
affected by liquidity problems in the TIPS market. However, some
market participants point out that the inflation swaps market is considerably more illiquid than the TIPS market, and inflation-swap
implied break-evens are consistently higher than TIPS-based breakevens by about 20bp.11 Therefore, it may be interesting to see if the
difference between inflation swap rates and TIPS inflation breakeven rates tends to be correlated with liquidity measures such as
the bidask spread in the TIPS market. Note that, during the market
dislocation in autumn 2008, bidask spreads widened considerably
in the TIPS market, but this effect was much more muted in the
inflation swap market.
Buraschi and Jiltsov (2005) provide a structural approach to estimating inflation expectations. They developed a real business cycle
model with a monetary channel to study the nature of deviations
from the expectations hypothesis and estimated a term structure of
the inflation risk premium. In their model, the inflation risk premium
was upward sloping, with the estimates between 20 and 140bp and
an average of 70bp at the 10-year maturity over the sample period
19612000.
An alternative model-free approach can be found in Grishchenko and Huang (2010), which, in the spirit of Evans (1998), is
arbitrage-free and also easy to implement. Specifically, this approach
takes the nominal and TIPS yields as given, and does not involve
any term structure model. Furthermore, various measures of inflation forecasts are used to identify expected inflation, and the inflation
risk premium can be estimated without using a term structure model
either. Grishchenko and Huang derived real yields by including
two explicit adjustments to TIPS yields, a three-month TIPS indexation lag correction and a liquidity premium. Taking these two into
227
i
i
i
Barr and Campbell (1997)
Model specification
Ang et al (2008)
Chernov and Mueller (2008)
DAmico et al (2009)
Grishchenko and Huang (2010)
Chen et al (2010)
Haubrich et al (2011)
Grishchenko et al (2011)
Data used
UK inflation-indexed bonds
N/A
US Treasury bonds, CPI,
real activity measures
US Treasury bonds,
CPI, M2
US Treasury bonds, CPI
US Treasury bonds,
CPI, GDP
US Treasury bonds,
TIPS, CPI
US Treasury bonds,
TIPS, CPI, surveys
US Treasury bonds, TIPS,
CPI, surveys
US Treasury bonds, TIPS, CPI
US Treasury bonds, TIPS,
surveys, inflation swap rates
US Treasury bonds, TIPS, CPI
Reference
INFLATION-SENSITIVE ASSETS
228
Table 10.2 Real and nominal term structure models
i
i
i
perrucci 2012/7/11 17:12 page 229 #251
account, they found that the inflation risk premium overall does not
exceed 10bp. This estimate is much lower than those obtained earlier via various structural model estimations. Table 10.2 provides a
summary of the models in this literature.
ZERO-COUPON TIPS YIELDS VERSUS REAL YIELDS
Liquidity effect in TIPS
Simple TIPS yields are biased estimates of real yields, as they are
known to contain a sizeable liquidity premium, at least in the early
years of the development of the TIPS market. Therefore, we need
to make an appropriate correction when working with TIPS yields.
Clearly, illiquidity drives TIPS prices down and TIPS yields up. If
we abstract for a moment from the issue of the indexation lag, the
R
difference between TIPS yields yt,TIPS
and real yields yt, is attributed
to a liquidity premium Lt,
R
yt,TIPS
= yt, + Lt,
EPt [It,inf
] + IRPt,
The above equation shows that the inflation risk premium might be
understated if the liquidity adjustment Lt, is ignored. We can use
several methods to estimate the liquidity premium. For example,
Lt, can be calculated as the difference between real yields (derived
within a term structure model) and actual TIPS yields. Alternatively,
we can use other estimation techniques that are less dependent on
model specification, such as simply comparing TIPS prices with a
benchmark fitted curve, or regressing TIPS break-evens on several
proxies for market liquidity.
As mentioned earlier, DAmico et al (2009) used a four-factor affine
term structure model to fit both nominal and real term structures.
In their model, the instantaneous real rate, the instantaneous nominal rate, the price of nominal rates risk and the price of real rates
risk are affine functions of three latent factors Xt = (X1t , X2t , X3t ).
In addition, the liquidity premium (and an instantaneous liquidity
229
i
i
i
perrucci 2012/7/11 17:12 page 230 #252
INFLATION-SENSITIVE ASSETS
spread) is explicitly modelled as the sum of a deterministic maturityindependent downwards trend Ld , and an (affine) stochastic component Lst, that depends on the three latent factors mentioned above
(ie, on the status of the economy) plus a fourth (TIPS-specific) factor
t orthogonal to the others
X
Lt, = Ld + Lst,
t
Lst, = a + bT Xt + c X
where bT is a 1 3 row vector and c is a scalar. The TIPS liquidity premium represents the difference between TIPS yields and real
yields and has both a deterministic component (which decreases the
premium over time to account for the increased liquidity in the US
TIPS market) and a stochastic component (in general, correlated with
the three latent factors driving both real and nominal yield curves).
DAmico et al found that the deterministic liquidity component premium was high in the early 2000s (as high as 120bp in 1999), but
came down significantly to about 10bp in 20045. After removing
this deterministic trend, DAmico et al found that stochastic component of the liquidity risk premium has been stationary since then
and had varied between 50 and 50bp. The term structure of the liquidity premium in their model is, generally, flat. DAmico et al also
found that the variation in the 10-year liquidity premium drives
over 20% (but with large standard errors) of the variation in inflation break-even rates, while the rest is due to variation in inflation
expectations (55%) and inflation risk premium (about 25%). As for
yields, US TIPS yield variance is dominated by changes in real yields
(119% at the 10-year horizon, with the rest 19% due to the liquidity
premium), while changes in the 10-year nominal yields are due to
changes in real yields (67%), changes in inflation expectations (23%)
and variation in inflation risk premium (10%), in contrast with, for
example, Ang et al (2008).
Pflueger and Viceira (2011) derive a higher estimate, about 70bp,
for the liquidity premium in normal years (outside early TIPS years,
when the liquidity spread was in excess of 100bp, and the 20089
financial crisis, when it reached over 200bp). They regress the TIPS
inflation break-evens on four proxies for market liquidity, ie, the offthe-run 10-year nominal spread, the GNMA spread to the on-the-run
nominal treasury, the difference in trading volumes between TIPS
and nominal treasuries, and the difference in 10-year asset-swap
230
i
i
i
perrucci 2012/7/11 17:12 page 231 #253
i
i
i
perrucci 2012/7/11 17:12 page 232 #254
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 233 #255
EPt [It,inf
] + IRPt,
i
i
i
perrucci 2012/7/11 17:12 page 234 #256
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 235 #257
1
1
Pt+ ,0 ] = Et [Nt+
1]
Nt1 Pt, = Et [Nt+
(10.1)
t
0
rs ds
(10.2)
the dynamics of the short rate rs determines the whole term structure,
ie, the entire zero-coupon yield curve yt,
Pt, = exp[yt, ] =
Q
Et
t+
exp
rs ds
t
(10.3)
235
i
i
i
perrucci 2012/7/11 17:12 page 236 #258
INFLATION-SENSITIVE ASSETS
(10.4)
(10.5)
t+
t+
t+
1
= exp
Ts s ds
Ts dBPs
2 t
t
t
(10.6)
t
0
s ds = Bt
(10.7)
Q
dBPt + t dt = dBt
(10.8)
BPt + t1 = Bt
(10.9)
Q
Et
t+
exp
rs ds
t
(10.10)
EPt
t+
t+
Mt+
exp
rs ds = EPt
t
Mt
t
(10.11)
236
i
i
i
perrucci 2012/7/11 17:12 page 237 #259
Mt = t exp
rs ds
(10.12)
(10.13)
EPt
1
exp
2
t+
t
Ts s
ds
t+
t
Ts
dBPs
t+
rs ds
exp
t
t+
t+
1
def P
= Et exp
Ts s ds
Ts dBPs
2 t
t
t+
P
rs ds
Et exp
t
t+
t+
1
P
T
T
P
+ covt exp
s s ds
s dBs ,
2 t
t
t+
exp
rs ds
t
(10.14)
where the last equality follows from the very definition of covariance. We define
t,
1
= exp
2
def
t+
t
Ts s
ds
t+
t
Ts dBPs
(10.15)
t+
exp
rs ds
t
t+
+ covPt (t, ), exp
rs ds
(10.16)
This equation can be used to calculate bond prices given the specification of the short rate and the price of risk under the physical
measure P.
237
i
i
i
perrucci 2012/7/11 17:12 page 238 #260
INFLATION-SENSITIVE ASSETS
(10.17)
The same result can be obtained working under the physical measure
P as follows
Pt,1 = EPt
t+1
exp(rt )
t
= exp(rt 2 Tt t ) exp(+ 2 Tt t )
= exp(rt )
(10.18)
1
rs ds
= yt, + ln EPt exp
t
Tt, =
ln(EPt [t, ])
1
ln 1 +
t+
rs ds)]
t t+
t
(10.19)
rs ds)]
(10.20)
EPt
t+1
exp(rt )Pt+1,1
t
(10.21)
238
i
i
i
perrucci 2012/7/11 17:12 page 239 #261
t+1
exp(rt )
t
(10.22)
i =1
mN
t+i
(10.23)
t,n =
Pt+n,n
1 P
Et ln
n
Pt,
=
1 P
Et ln[Pt+n,n ] + yt,
n
n
(10.25)
1 P
Et ln[Pt+n,n ] + yt, yt,n
n
n
(10.26)
(10.27)
(10.28)
(10.29)
(10.30)
239
i
i
i
perrucci 2012/7/11 17:12 page 240 #262
INFLATION-SENSITIVE ASSETS
n =
and n = n
(10.31)
(10.33)
(10.34)
TX Xt
(10.35)
t = 0 + X Xt
(10.36)
rt = 0 +
We will work in discrete time in order to deal with finite difference equations rather than differential equations. Introduced in the
appendix on page 238, the following relation
Pt, =
EPt
t+1
exp(rt )Pt+1,1
t
(10.37)
can help us build the whole yield curve at time t, starting from
the one-period short rate rt . Furthermore, since this rate is assumed
to be affine in the state vector, this property clearly extends to the
whole yield curve, thanks to the recursive expression above. In fact,
suppose Pt,1 is exponentially affine, ie
Pt,1 = exp(A1 + BT1 Xt )
(10.38)
240
i
i
i
perrucci 2012/7/11 17:12 page 241 #263
t+1
exp(rt )Pt+1,1
t
= exp(rt 12 Tt t + A1 )
EPt [exp(Tt BP1 + BT1 Xt+1 )]
= exp(rt 12 Tt t + A1 )
EPt [exp(Tt BP1 + BT1 Xt+1 )]
1
= exp(rt 2 Tt t + A1 + BT1 + BT1 Xt )
(10.39)
where the following relation has been used to arrive at the last
equality
Xt+1 = + Xt + BPt+1
(10.40)
(10.41)
Note the quadratic terms in the price of risk cancel out, so we are
left with
Pt, = exp(rt + A1 + BT1 + BT1 Xt
+ 12 BT1 T B1 BT1 t )
(10.42)
(10.43)
(10.44)
BT
(10.45)
TX
BT1 (
X )
i
i
i
perrucci 2012/7/11 17:12 page 242 #264
INFLATION-SENSITIVE ASSETS
Pt,R = Et
t+
exp
Pt,R = EPt
rsR ds
MtR+
MtR
= EPt
t+
tR+
R
exp
r
ds
s
tR
t
(10.46)
= [exp(yt,R )]
QR
dBPt + Rt dt = dBt
(10.47)
(10.48)
QN
1
1
Nt1 Qt Pt,R = Et [Nt+
Qt+ PtR+ ,0 ] = Et [Nt+
Qt+ ]
(10.49)
EPt
t+
tN+
Qt+
N
exp
r
ds
s
Qt
tN
t
N
Mt+ Qt+
MtN Qt
MtR+
MtR
(10.50)
(10.51)
242
i
i
i
perrucci 2012/7/11 17:12 page 243 #265
Qt+
Qt0
(10.52)
=
=
QN
Et
t+
Qt+
N
exp
rs ds
Qt QN
E
Qt0 t
Qt
Qt0
Qt
Qt0
Qt
t+
Qt+
N
exp
rs ds
t
t+
tN+
EPt
exp
rsN
tN
t
N
Mt+ Qt+
EPt
MtN Qt
Qt
ds
Qt+
Qt
(10.53)
Qt R
P
Qt0 t,
where
(10.54)
QN
dBPt + N
t dt = dBt
(10.55)
PtTIPS
=
0 ,t+ ,0
Qt+3M
Qt+3M
+ max 1
,0
Qbase
Qbase
(10.56)
where Qbase is the index base value (see Chapter 7). Typically, the
value of the deflation floor is neglected when computing zerocoupon TIPS yields. This introduces a negative bias to real yields calculated without including the floor, although it could be argued that,
243
i
i
i
perrucci 2012/7/11 17:12 page 244 #266
INFLATION-SENSITIVE ASSETS
MtN+ Qt+3M
MtN Qbase
=
PtTIPS
0 ,t,
Qt
MtN+
Pt,R3M EPt
Qbase
MtN+3M
+ covPt
Qt R
MtN+
Pt,3M EPt
Qbase
MtN+3M
1 +
covPt
If we define
lagt, =
ln EPt
(10.57)
MtN+
MtN+3M
covPt
1
ln
1+
(10.58)
(10.59)
and since (see Chapter 7)
Qbase TIPS
TIPS
P
= exp[yt,x
]
Qt t0 ,t,
(10.60)
we then have
R
yt,TIPS
= yt,3M + lagt,
(10.61)
yt,R3M = yt,TIPS
lagt,
(10.62)
t+
t
is ds + QT (BPt+ BPt )
(10.63)
244
i
i
i
perrucci 2012/7/11 17:12 page 245 #267
(10.64)
d ln Qt = it dt + QT dBPt
(10.65)
or, alternatively
Note that
Qt
=
ln
Qt+
and
EPt
t+
t
is ds QT (BPt+ BPt )
Qt
ln
Qt+
=
EPt
(10.66)
t+
t
is ds
(10.67)
t+
rsR ds
t+
+ covPt Rt, , exp
rsR ds
t
t+
P
R
P
R
rs ds
= Et [t, ]Et exp
t
t+
covPt [Rt, , exp( t rsR ds)]
1+ P R
t+
Et [t, ]EPt [exp( t rsR ds)]
(10.68)
245
i
i
i
perrucci 2012/7/11 17:12 page 246 #268
INFLATION-SENSITIVE ASSETS
where Rt, is a function of the price of real risk, similar to that defined
previously. If we introduce a real yield term premium as
Tt,R =
ln(EPt [Rt, ])
ln 1 +
t+
rsR ds)]
t+
t
(10.69)
rsR ds)]
1
yt,R = EPt exp
rsR ds
+ Tt,R
(10.70)
t+
t
rsR ds
(10.71)
EPt (It,R ) +
ln EPt [exp(It,R )] +
Jensens term
(10.72)
If we assume It,R is normally distributed conditional to information at time t (ie, an affine model), then we can explicitly calculate
the Jensen term and we have
EPt [exp(It,R )] = exp[EPt (It,R ) + 12 vart (It,R )]
ln EPt [exp(It,R )] = EPt (It,R ) + 12 vart (It,R )
Therefore
yt,R =
EPt (It,R )
1
vart (It,R ) + Tt,R
2
(10.73)
(10.74)
(10.75)
where the first term on the right-hand side represents the -period
expectation of the instantaneous real rate at time t, the second term
is the Jensen inequality term and the last term is the real rates term
premium.
Nominal yields and the inflation risk premium
We have already derived nominal yields and term premium in a
previous appendix, but here we express these in terms of real yields,
246
i
i
i
perrucci 2012/7/11 17:12 page 247 #269
MtN+
MtN
= EPt
MtR+ Qt
MtR Qt+
MtR+ P Qt
Et
Qt+
MtR
= Pt,R EPt
= Pt,R EPt
yt,N = yt,R
Qt
Qt+
Qt
Qt+
EPt
+ covPt
1+
Qt
Qt+
+ covPt
MtR+ Qt
,
MtR Qt+
MtR+ Qt
,
MtR Qt+
ln 1 +
(10.76)
Qt
Qt+
Qt
Qt
Qt
= EPt ln
+ ln EPt
EPt ln
Qt+
Qt+
Qt+
Jensens term
(10.78)
The inflation risk premium, IRPt, , is given by
IRPt, =
ln 1 +
(10.79)
Qt+
(10.80)
we have
yt,N = yt,R +
EPt
t+
t
ln EPt
is ds
Qt
Qt+
EPt
Qt
ln
Qt+
+ IRPt,
(10.81)
Jensens term
247
i
i
i
perrucci 2012/7/11 17:12 page 248 #270
INFLATION-SENSITIVE ASSETS
Collecting the two (real rate and price index) Jensens terms
together
1
1
Jensent, = EPt [exp(It,R )] EPt (It,R )
ln EPt
Qt
Qt+
Qt
1
+ EPt
ln
Qt+
(10.82)
so that
yt,N =
1
EPt (It,R ) +
EPt [It,inf
]
expectations
Tt,R
IRPt,
+ Jensent,
(10.83)
inflation premium
(10.84)
QN
BPt + N
t dt = dBt
(10.85)
(10.86)
N
t
(10.87)
N
0
NT
X Xt
(10.88)
QR
BPt + Rt dt = dBt
(10.89)
(10.90)
Rt
(10.91)
R0
RT
X Xt
t+
t
is ds + QT (BPt+ BPt )
(10.92)
248
i
i
i
perrucci 2012/7/11 17:12 page 249 #271
(10.93)
(10.94)
(10.95)
(10.96)
P
T
T
P
N
rtR dt RT
t dBt = (it + 2 Q Q ) dt + Q dBt rt dt
1
P
NT
NT
t dBt t Q dt
(10.97)
so that
rtR = rtN (it + 12 QT Q ) + NT
t Q
(10.98)
Rt
(10.99)
N
t
Thus,
i
T
NT
R0 = N
0 0 2 Q Q + 0 Q
(10.100)
RX
R0
RX
NT
X Q
(10.101)
=
=
=
N
X
N
0
N
X
iX
(10.102)
(10.103)
Note that we have been working under the data-generating probability measure, where the price index satisfies
dQt
1
= (it + 2 QT Q ) dt + QT dBPt
Qt
(10.104)
249
i
i
i
perrucci 2012/7/11 17:12 page 250 #272
INFLATION-SENSITIVE ASSETS
dBPt = dBt
N
t dt = dBt
(10.105)
Rt dt
(10.106)
we can derive the price index evolution under the two (nominal and
real) risk-neutral measures
dQt
QN
= (rtN rtR ) dt + QT dBt
Qt
QR
(10.107)
Equivalent probability measures agree on what is possible. In other words, impossible events
will have zero probability under both measures, and possible events will have different but
positive probabilities for both measures.
In particular, only a number of yields equal to the number of latent factors can be exactly fitted
at all points in time (for example, three yields can be exactly fitted in a three-latent-variable
model).
Latent factors refer to their lack of further interpretation, although there are cases where a
macroeconomic identification might be possible.
In other words, the ideal model should be flexible enough to capture the term-structure
dynamics, and yet remain mathematically tractable.
Note that in many older treatments of this topic, real yields were assumed constant, and
therefore nominal yields changes were simply driven by changes in inflation expectations
(the risk premium is obviously zero in this case). This is partly due to the fact that real yields
are observable (lags and liquidity premium consideration aside) only in countries where
a government inflation-linked market is developed, so the empirical study of real yields
dynamics has in many cases been limited by the availability of data.
In a later paper (Ang et al 2005), the Taylor rule includes only one latent factor contemporaneously uncorrelated with growth and inflation. The latter assumption is reasonable (as growth
and inflation will react with a lag to monetary shocks), and allows for unbiased (albeit not
efficient) ordinary least square estimation of the coefficients in the Taylor equation.
Grishchenko and Huang (2010) estimate that this effect does not exceed four basis points in
terms of the real yield. In addition, inflation-linked bonds also have duration risk.
Here and later, we refer to real term structure models as models that include the dynamics of the instantaneous real rate and the instantaneous inflation, expected inflation or the
instantaneous nominal rate.
To be precise, there is both a risk premium (covariance term) and a Jensens inequality term
(variance term).
250
i
i
i
perrucci 2012/7/11 17:12 page 251 #273
REFERENCES
Adrian, T., and H. Wu, 2010, The Term Structure of Inflation Expectations, Staff
Report 362, Federal Reserve Bank of New York.
Ahn, D.-H., R. Dittmar and R. A. Gallant, 2002, Quadratic Term Structure Models:
Theory and Evidence, Review of Financial Studies 15, pp. 24388.
Ang, A., G. Bekaert and M. Wei, 2008, The Term Structure of Real Rates and Expected
Inflation, The Journal of Finance 63, pp. 797849.
Ang, A., and M. Piazzesi, 2003, A No-Arbitrage Vector Autoregression of Term Structure
Dynamics with Macroeconomic and Latent Variables, Journal of Monetary Economics 50,
pp. 74587.
Barnes, M., Z. Bodie, R. Triest and J. C. Wang, 2010, A TIPS Scorecard: Are They
Accomplishing Their Objectives?, Financial Analysts Journal 66, pp. 6884.
Barr, D. G., and J. Y. Campbell, 1997, Inflation, Real Interest Rates, and the Bond Market:
A Study of UK Nominal and Index-Linked Government Bond Prices, Journal of Monetary
Economics 39, pp. 36183.
Bekaert, G., and X. S. Wang, 2010, Inflation Risk and the Inflation Risk Premium, SSRN
eLibrary, URL: http://ssrn.com/abstract=1600312.
Bikbov, R., and M. Chernov, 2008, Monetary Policy Regimes and the Term Structure of
Interest Rates, CEPR Discussion Paper DP7096, URL: http://ssrn.com/abstract=1344675.
Bitsberger, T. S., 2003, Inflation Indexing of Government Securities, Presentation to the
Bond Market Associations Inflation-Linked Securiries Conference, JS-505, URL: http://
treasury.gov/.
Buraschi, A., and A. Jiltsov, 2005, Inflation Risk Premia and the Expectation Hypothesis,
Journal of Financial Economics 75, pp. 42990.
Campbell, J., and R. Shiller, 1991, Yield Spreads and Interest Rate Movements: A Birds
Eye View, Review of Economic Studies 58, pp. 495514.
Campbell, J. Y., and L. M. Viceira, 2001, Who Should Buy Long-Term Bonds?, American
Economic Review 91, pp. 99127.
Campbell, J., L. Chan and L. M. Viceira, 2003, Strategic Asset Allocation in a ContinuousTime VAR Model, Journal of Financial Economics 67, pp. 79120.
Campbell, J., R. Shiller and L. M. Viceira, 2009, Understanding Inflation-Indexed Bond
Markets, Brookings Papers on Economic Activity, Spring, pp. 79120.
Chen, R.-R., B. Liu and X. Cheng, 2010, Pricing the Term Structure of Inflation Risk
Premia: Theory and Evidence from TIPS, Journal of Empirical Finance 17, pp. 70221.
Chernov, M., and P. Mueller, Forthcoming, The Term Structure of Inflation Expectations,
Journal of Financial Economics.
Cox, J., J. Ingersoll and S. Ross, 1985, A Theory of the Term Structure of Interest Rates,
Econometrica 53, pp. 385408.
251
i
i
i
perrucci 2012/7/11 17:12 page 252 #274
INFLATION-SENSITIVE ASSETS
Dai, Q., and K. Singleton, 2000, Specification Analysis of Affine Term Structure Models,
The Journal of Finance 55, pp. 194378.
Dai, Q., and K. Singleton, 2002, Expectations Puzzles, Time-Varying Risk Premia, and
Dynamic Models of Term Structure, Journal of Financial Economics 63, pp. 41541.
Dai, Q., A. Le and K. Singleton, 2010, Discrete-Time Affine Q Term Structure Models
with Generalized Market Prices of Risk, Review of Financial Studies 23, pp. 2184227.
DAmico, S., D. H. Kim and M. Wei, 2009, Tips from TIPS: The Informational Content
of Treasury Inflation-Protected Security Prices, Working Paper 2010-19, Federal Reserve
Board, Washington, DC.
Dudley, W., J. Roush and M. Steinberg Ezer, 2009, The Case for TIPS: An Examination
of the Costs and Benefits, Economic Policy Review 15(1).
Duffee, G., Forthcoming, The Term Structure and the Macroeconomy, in G. Constantinides, M. Harris and R. Stultz (eds), Handbook of the Economics of Finance (Elsevier).
Duffie, D., and R. Kan, 1996, AYield-Factor Model of Interest Rates, Mathematical Finance
6, pp. 379406.
Engle, R., 2002, Dynamic Conditional Correlation: A Simple Class of Multivariate Generalized Autoregressive Conditional Heteroskedasticity Models, Journal of Business and
Economic Statistics 20, pp. 33950.
Evans, M., 1998, Real Rates, Expected Inflation, and Inflation Risk Premia, The Journal
of Finance 53, pp. 187218.
Fama, E. F., 1990, Term-Structure Forecasts of Interest Rates, Inflation and Real Returns,
Journal of Monetary Economics 25, pp. 5676.
Fama, E. F., and R. R. Bliss, 1987, The Information in Long-Maturity Forward Rates, The
American Economic Review 77, pp. 68092.
Fleckenstein, M., F. A. Longstaff and H. Lustig, 2010, Why Does the Treasury Issue TIPS?
The TIPS-Treasury Bond Puzzle, Working Paper, UCLA.
Grishchenko, O. V., and J.-Z. Huang, 2010, Inflation Risk Premium: Evidence from the
TIPS Market, SSRN eLibrary, URL: http://ssrn.com/abstract=1108401.
Grishchenko, O. V., J. M. Vanden and J. Zhang, 2011, The Informational Content of the
Embedded Deflation Option in TIPS, SSRN eLibrary, URL: http://ssrn.com/abstract=
1695511.
Gurkaynak, R., B. Sack and J. H. Wright, 2010, The TIPS Yield Curve and Inflation
Compensation, American Economic Journal: Macroeconomics 2, pp. 7092.
Haubrich, J., G. G. Pennacchi and P. Ritchken, 2011, Estimating Real and Nominal Term
Structures Using Treasury Yields, Inflation Forecasts, and Inflation Swap Rates, SSRN
eLibrary, URL: http://ssrn.com/abstract=1361219.
Heath, D., R. Jarrow and A. Morton, 1992, Bond Pricing and the Term Structure of Interest
Rates: A New Methodology, Econometrica 60, pp. 77105.
Ho, H.-W., H. Huang and Y. Yildirim, 2011, Affine Model of Inflation-Indexed Derivatives
and Inflation Risk Premium, Working Paper, National Central University of Taiwan and
Syracuse University.
Ho, T. S. Y., and S.-B. Lee, 1986, Term Structure Movements and Pricing Interest Rate
Contingent Claims, The Journal of Finance 41, pp. 101128.
Hrdahl, P., and O. Tristani, 2007, Inflation Risk Premia in the Term Structure of Interest
Rates, Working Paper Series, No. 734, European Central Bank.
252
i
i
i
perrucci 2012/7/11 17:12 page 253 #275
Huang, J.-Z., and Z. Zhong, 2011, Time Variation in Diversification Benefits of Commodity, REITs, and TIPS, Journal of Real Estate Finance and Economics, DOI:10.1007/
s11146-011-9311-6.
Kim, D. H., 2009, Challenges in Macro-Finance Modeling, Federal Reserve Bank of St Louis
Review, September/October, pp. 51944.
Kothari, S. P., and J. Shanken, 2004, Asset Allocation with Inflation-Protected Bonds,
Financial Analysts Journal 60, pp. 5470.
Leippold, M., and L. Wu, 2002, Option Pricing under the Quadratic Class, Journal of
Financial and Quantitative Analysis 37, pp. 27195.
Litterman, R., and J. Scheinkman, 1991, Common Factors Affecting Bond Returns,
Journal of Fixed Income 1, pp. 5461.
Morningstar, 2010, Fund Flows Update, Report, February.
Nelson, C. R., and A. F. Siegel, 1987, Parsimonious Modeling of the Yield Curves, Journal
of Business 60, 47389.
Pflueger, C., and L. M. Viceira, 2011, An Empirical Decomposition of Risk and Liquidity
in Nominal and Inflation-Indexed Government Bonds, Working Paper 11-094, Harvard
Business School.
Piazzesi, M., 2010, Affine Term Structure Models, in Y. Ait-Sahalia and L. Hansen (eds),
Handbook of Financial Econometrics (Amsterdam: North-Holland).
Roll, R., 2004, Empirical TIPS, Financial Analysts Journal 60, pp. 3153.
Roush, J., 2008, The Growing Pains of TIPS Issuance, Finance and Economics
Discussion Series Working Paper 2008-08, Federal Reserve Board, Washington, DC.
Singleton, K. (2006), Empirical Dynamic Asset Pricing (Princeton University Press).
Svensson, L. E. O., 1994, Estimating and Interpreting Forward Rates: Sweden 19921994,
National Bureau of Economic Research Working Paper 4871.
Vasicek, O., 1977, An Equilibrium Characterization of the Term Structure, Journal of
Financial Economics 5, pp. 17788.
253
i
i
i
perrucci 2012/7/11 17:12 page 254 #276
i
i
i
perrucci 2012/7/11 17:12 page 255 #277
11
i
i
i
perrucci 2012/7/11 17:12 page 256 #278
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 257 #279
i
i
i
perrucci 2012/7/11 17:12 page 258 #280
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 259 #281
i
i
i
perrucci 2012/7/11 17:12 page 260 #282
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 261 #283
prices are not of great concern. The credibility afforded to monetary policy by inflation targeting can also diminish economies vulnerability to the inflationary impact of commodity price shocks,
by ensuring those shocks do not become embedded in expected
inflation.
The narrow focus on the real economy and performance associated with inflation targeting nevertheless means that important
information concerning developments in money and financial market variables may have been ignored in setting monetary policy.
Moreover, inflation targeting operates through one policy variable:
the central bank policy interest rate. Whether that interest rate is a
sufficiently robust instrument in avoiding or addressing major disturbances to economic and financial performance is open to question. Taylor (2009) apportions part of the blame for the financial crisis
that emerged in 2007 to monetary policy having pursued persistently
low interest rates for an extended period. He suggests that monetary excesses were the main cause of the boom and subsequent
bust that occurred in the mid 2000s (Taylor 2009, p. 2).
While inflation targets were being met in most developed countries during the mid 2000s, monetary policy, mainly through the
provision of cheap credit, was contributing to excessive risk-taking
in financial markets, and an over-pricing of financial and real assets
such as property. Eventually, this led to substantial falls in asset
prices, as well as the threat of deflation hanging over Western
economies. The inflation-targeting framework could not have foreseen this, because of its belief in the second-order importance of asset
markets for monetary policy.
Besides raising potential issues for the maintenance of price stability, an accommodative monetary policy stance, if maintained over
a long period, can pose a threat to financial stability. This view that
financial imbalances, especially excess money and credit growth,
brought about by monetary policy pose a threat to the well-being
of economies was most prominently expressed by Bank for International Settlements (BIS) economists (see, for example, Borio and
Lowe 2002; Borio and White 2004) in the years leading up to the
financial crisis. They argue that the simultaneous development of
imbalances in monetary variables, such as credit, and in asset prices
should be of concern to central banks. Financial liberalisation and
innovation can generate such imbalances and encourage greater
261
i
i
i
perrucci 2012/7/11 17:12 page 262 #284
INFLATION-SENSITIVE ASSETS
procyclicality in financial markets. This can occur against a background of low and stable inflation. Excess demand can show up
first in asset prices rather than consumer prices, which may explain
why financial imbalances and rising asset prices occur in a lowinflation environment. The concerns expressed by BIS and other
economists underline the need for central banks to monitor monetary and financial developments, as well as those in the real economy,
closely.
MONETARY POLICY IN THE FINANCIAL CRISIS
Inflation targeting, which seemed to do a good job in the not too
distant past, has been found wanting. As others have noted (see, for
example, Canuto 2009), well-behaved inflation and output performance, which were features of advanced industrial economies leading up to the crash, are no guarantee against a dangerous upward
asset price spiral developing and then collapsing, with enormous
implications for the central bank and its ability to maintain a stable
monetary and financial system.
In the wake of the financial crisis, central banks, faced with either a
liquidity trap or the zero lower bound on nominal interest rates, had
to turn to using an alternative policy instrument, quantitative easing.
This was utilised because nominal interest rates were already close
to zero, and thus the scope for reducing them further was limited.
Quantitative easing involves proactively buying up a large fraction
of the stock of government bonds, at whatever price is needed for
holders to be willing to engage in exchange. Its purpose is to accommodate the need for liquidity in financial markets, and the economy
more generally.
It goes beyond, however, a normal accommodating monetary policy, which tends to occur at a positive value for the nominal rate of
interest, when the central bank makes funding available in infinitely
elastic amounts at that rate (subject to good collateral). The difficulty
with conventional monetary policy is that the amount of liquidity
injected into the financial system may be deemed to be inadequate,
even with full accommodation and at a zero rate of interest, in the
type of distressed state in which financial markets found themselves
in the wake of the financial crisis. This is where the need for quantitative easing comes in. It involves purchasing what has turned
262
i
i
i
perrucci 2012/7/11 17:12 page 263 #285
out to be an extremely large amount of government securities outright, as a way of injecting liquidity into the banking system, with
the ultimate objective of kick-starting the economy, and obviating
deflation.
Quantitative easing was employed by the Bank of Japan during
the early-to-mid 2000s in its attempts to tackle deflation in the
Japanese economy. Assessments of the impact of this policy are that
it had limited effect in raising aggregate demand and prices, but provided some support to the countrys banking sector (Spiegel 2006;
Ugai 2007).
The Federal Reserve and other central banks have introduced
extremely large amounts of funds into the financial system through
quantitative easing. The sizes of their balance sheets have increased
considerably since 2007. This expansion of liquidity was confined
to the banking sector initially. It has since resulted in a corresponding improvement in the supply of credit and liquidity to the retail
non-banking sector.
Quantitative easing might also affect behaviour in financial markets in another way. Large purchases of government securities by the
central bank are likely to drive their prices to levels that reduce, if
not eliminate altogether, their attractiveness as an investment option.
Koo (2011) argues that, with the private sector deleveraging due to
balance-sheet difficulties, fund managers, devoid of both private
sector and public sector borrowers, will turn to commodities as an
alternative investment option. This effect may have been at play in
commodity price behaviour since the introduction of programmes
of quantitative easing.
It is important to remember that this is a portfolio-rebalancing
effect and may not have a lasting effect on commodity prices, given
that other asset markets should be expected to return to normality
at some time in the future. The portfolio-rebalancing effect, however, is distortionary in the short run, as investors are effectively
constrained into purchasing commodities. This cannot be beneficial
to commodity markets, and to the efficient allocation of investment
resources more generally within the economy.
The acceleration in the growth of the M2 money stock in the latter
half of 2008 while the US economy was still in recession is likely
to have helped maintain momentum to commodity prices.4 In the
next section, we discuss how a monetary shock has a proportionate
263
i
i
i
perrucci 2012/7/11 17:12 page 264 #286
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 265 #287
i
i
i
perrucci 2012/7/11 17:12 page 266 #288
INFLATION-SENSITIVE ASSETS
real interest rate for commodities is that it represents the opportunity cost of holding them. A rise in the real interest rate then reduces
the demand for commodities, causing their real prices to fall. In this
way, monetary policy has an impact on commodity prices through
its effect on real interest rates.
In the Frankel model, the amount by which commodity prices
decline is determined by a no-arbitrage condition. Commodity
prices must fall to the extent that their subsequent appreciation to
long-run values compensates their holders fully for the increased
cost of carrying them. Prices, then, overshoot equilibrium values
to meet this market requirement. Subsequently, the CPI inflation
rate will itself adjust (slowly) upwards, and the real interest rate
will decline, acting to restore equilibrium to the commodity market.
In his 2008 article, Frankel emphasises the relevance of his overshooting theory to developments in commodity prices around that
time. He attributes the rise in commodity prices in 20024 to declining real interest rates. A number of other studies find a similar relationship between commodity prices and real interest rates. Using
quarterly data covering the years 19902007, Akram (2009) finds
commodity prices increase significantly in response to reductions in
real interest rates. The econometric results of Anzuini et al (2010)
indicate that expansionary US monetary policy shocks increase
commodity prices, albeit to a limited extent.
In Browne and Cronin (2010), we also provide an overshooting theory of commodity prices. Whereas Frankels perspective
draws on the Dornbusch (1976) theory of exchange rate overshooting in framing his model, we use two (essentially Friedman-style)
monetarist propositions to develop ours. These are that exogenous
changes in the nominal money stock lead to equivalent percentage changes in the overall price level (comprising commodity and
consumer good prices), and that exogenous changes in-the-money
stock are neutral in the long run, implying that all individual prices,
whether they be of consumer goods or commodities, adjust over
time in the same proportion as the money stock, thus leaving all
relative prices unchanged.
When these two propositions are combined with an acknowledgement that commodity prices are more flexible than consumer prices,
commodity prices are shown to overshoot their new long-run equilibrium values in response to a change in the exogenous money
266
i
i
i
perrucci 2012/7/11 17:12 page 267 #289
i
i
i
perrucci 2012/7/11 17:12 page 268 #290
INFLATION-SENSITIVE ASSETS
negative shock)
Mt = (1 + )Mt1
Pt =
Mt Vt
Yt
The overall price level also rises by the same percentage amount,
given the proposition that it moves contemporaneously with the
size of the money stock (Vt1 = Vt = 1; Yt1 = Yt = Y). Without
other shocks, the new equilibrium overall price level PNEW EQ will
hold indefinitely into the future
Pt = (1 + )Pt1 = PNEW EQ = wFNEW EQ + (1 w)SNEW EQ
Over time, both commodity and consumer price indexes converge to
their equilibrium levels FNEW EQ and SNEW EQ , but it is what happens
in the meantime that is of most interest. In particular, we assume that
the consumer price index is sticky for one period after the money
shock, that is
St = (1 + )St1 with <
while commodity prices, which are traded on spot auction markets,
are fully flexible, and move percentage points in order to maintain
overall price equilibrium, ie
Pt = (1 +)Pt1 = wFt +(1 w)St = w(1 +)Ft1 +(1 w)(1 + )St1
or, rearranging the terms above
1 w St1
1+
1+
=1+
1
1+
w Ft1
1+
i
i
i
perrucci 2012/7/11 17:12 page 269 #291
so that
St+1 St =
w
[Ft FNEW EQ ] = [St SNEW EQ ]
1w
i
i
i
perrucci 2012/7/11 17:12 page 270 #292
INFLATION-SENSITIVE ASSETS
SA
US Department of Labor:
Bureau of Labor
Statistics
CRBSI
Index:
1967 = 100
NSA
Commodity
Research Bureau
M2
US$ billion
SA
Real GDP
Billions of
SAAR US Department of
chained 2005
Commerce: Bureau of
US dollars
Economic Analysis
0.03
M2
CPI
GDP
0.02
0.01
0
0.01
i
i
i
perrucci 2012/7/11 17:12 page 271 #293
i
i
i
perrucci 2012/7/11 17:12 page 272 #294
INFLATION-SENSITIVE ASSETS
40
20
0
Q3 2009
Q1 2004
Q3 1998
Q1 1993
Q3 1987
Q1 1982
Q3 1976
Q1 1971
40
Q3 1965
20
Q1 1960
60
16
14
12
10
8
6
4
2
0
2
4
Figure 11.2 Year-on-year CPI inflation rates and commodity price gap,
Q1 1960Q1 2011 in percentage points
CRBSI
CPI
Q1 2001
Q3 2001
Q1 2002
Q3 2002
Q1 2003
Q3 2003
Q1 2004
Q3 2004
Q1 2005
Q3 2005
Q1 2006
Q3 2006
Q1 2007
Q3 2007
Q1 2008
Q3 2008
Q1 2009
Q3 2009
Q1 2010
Q3 2010
Q1 2011
40
30
20
10
0
10
20
30
40
i
i
i
perrucci 2012/7/11 17:12 page 273 #295
Q2 2008. With the rate of change in the CPI not keeping pace with
that in M2, the monetary impulse affected commodity prices.
The differential between the M2 growth rate and the CPI inflation
rate is particularly noticeable between 2001 and 2003, as can be seen
in Figure 11.3. The rate of change in the CRBSI did not pick up until
2003 and was maintained into 2005. The overshooting theory would
explain this by noting that the response of commodity prices to a
monetary stimulus is not instantaneous, but rather occurs with a
lag; however, that lag is relatively short and the response is faster
than that of consumer prices.
There was some pick-up in the CPI inflation rate in 20045, as
the rate of commodity price inflation declined. This would be consistent with the eventual catch-up in CPI inflation rates that would
be expected in the wake of strong money growth, and in the initial
momentum to commodity prices from that monetary source falling
away. The period from early 2006 to mid 2008 saw a fresh surge in
commodity prices taking place. Money growth was also above the
CPI inflation rate at that time. As in 2005, the decline in the rate of
increase in commodity prices that was occurring just prior to Q3 2008
coincided with a steady rise in the CPI inflation rate to a value of 5%
in Q2 2008.
The vertical line at Q3 2008 in Figure 11.3 marks the start of a
sudden collapse in commodity prices in the second half of that year.
Year-on-year rates of growth in the CRBSI remained negative until
Q1 2010. Given that the rate of money growth was much greater
than the rate of the CPI inflation rate in late 2008 and through
2009 when commodity prices were, in general, falling, this commodity price behaviour may appear unusual.7 We suggest that a
flight from risky assets, such as commodities, to safe haven assets,
like money (accommodated by monetary policy), by consumers and
investors was the predominant force at work in financial markets in
late 2008/early 2009 and goes a long way towards explaining the
money and commodity price growth rates in Figure 11.3 during
that time. Since the level of uncertainty and investor nervousness
in the economy has receded somewhat since then, it is unsurprising that strong money growth manifests itself in rising commodity
prices. A positive commodity price gap is now evident (Figure 11.2).
This raises the prospect of CPI inflation rates rising in the years
ahead.
273
i
i
i
perrucci 2012/7/11 17:12 page 274 #296
INFLATION-SENSITIVE ASSETS
CONCLUSION
In this chapter, we described how inflation targeting became a central plank of modern monetary policy. It contributed to a reduction in inflation rates from the high values that prevailed in the
1970s and 1980s, and did so mainly through succeeding in anchoring inflation expectations close to low levels. Inflation targetings
biggest shortcoming, in our view, is that it has not taken account
of the vast changes that have occurred in the financial system since
the 1990s. These changes have altered the patterns of substitution
between money and financial assets. In our view, these new patterns are an important ingredient of the boombust cycles of the
1990s and 2000s. They have had the effect of enhancing the role of
money in the economy, but arguably in a disruptive way: something which was at the heart of the 20079 financial crisis, but
which most advocates of inflation targeting regard as something
of a sideshow.
Our overshooting model provides some input into understanding the potential for money to affect prices, including relative prices
between different classes of goods, such as commodities and consumer goods. It shows that monetary shocks can cause commodity
prices to react quickly to maintain equilibrium in the overall price
level. This can arguably prove unsettling for investors, as rising commodity prices may be interpreted as signalling a higher sustained
level of real demand for a commodity or commodity class when,
in fact, their prices are only reacting to a generalised, ie, monetary,
stimulus to prices in the economy. Monetary policymakers, mistakenly reading rising commodity prices as caused by market-specific
demand or supply shocks, might thus adopt inappropriate monetary policy responses, including no response at all. For both investors
and central banks, money is a variable that needs to be analysed and
understood, and not neglected.
The views expressed in this chapter are those of the authors and do
not necessarily represent the views of the Central Bank of Ireland
or the European System of Central Banks. We would like to thank
the editors for their very helpful comments and suggestions.
1
Using an econometric technique called band-pass filtering, Cuddington and Jerrett (2008)
provide evidence consistent with there having been three super-cycles in metal prices since
around the middle of the 19th century, with world metal markets currently being in the
early stages of a fourth super-cycle. They note that the latter is being attributed to Chinese
urbanisation and industrialisation.
274
i
i
i
perrucci 2012/7/11 17:12 page 275 #297
The quantity theory of money links money supply with the overall price level.
Uncovered by the seminal work of A. W. Phillips, the original Phillips curve refers to the
inverse relationship between unemployment and money wage rates. Other versions of the
model were later introduced, linking inflation with various measures of broad economic
activity and, later, inflation expectations.
Within the CPI, there are goods whose prices are more flexible than others. Bils and Klenow
(2004) find the fresh-food, energy-related products and durable-goods components of the CPI
to change relatively frequently. In the euro area, energy and unprocessed food have the most
flexible prices among consumer goods, while services have the lowest (lvarez et al 2006).
See, for example, Webb (1988), Garner (1989), Marquis and Cunningham (1990), Cody and
Mills (1991), Pecchenino (1992), Blomberg and Harris (1995) and Furlong and Ingenito (1996).
Another factor at play here is that the massive amount of hoarding of liquid balances induced
by the crisis has undermined the assumption of constant velocity, as often happens in
recessions.
REFERENCES
Akram, Q. F., 2009, Commodity Prices, Interest Rates and the Dollar, Energy Economics
31(6), pp. 83851.
lvarez, L., E. Dhyne, M. Hoeberichts, C. Kwapil, H. Le Bihan, P. Lnnemann, F. Martins,
R. Sabbatini, H. Stahl, P. Vermeulen and J. Vilmunen, 2006, Sticky Prices in the Euro
Area: A Summary of New Micro Evidence, Journal of European Economic Association 4,
pp. 57584.
Anzuini, A., M. Lombardi and P. Pagano, 2010, The Impact of Monetary Policy Shocks
on Commodity Prices, Working Paper 1232, European Central Bank.
Barsky, R. B., and L. Kilian, 2002, Do We Really Know that Oil Caused the Great Stagflation? A Monetary Alternative, in B. Bernanke and K. Rogoff (eds), NBER Macroeconomics
Annual 2001, pp. 13783.
Bils, M., and P. Klenow, 2004, Some Evidence on the Importance of Sticky Prices, Journal
of Political Economy 112, pp. 94785.
Blanchard, O., and J. Gal, 2010, The Macroeconomic Effects of Oil Price Shocks: Why
Are the 2000s So Different from the 1970s?, in J. Gal and M. Gertler (eds), International
Dimensions of Monetary Policy, pp. 373428 (Chicago, IL: University of Chicago Press).
Blomberg, S. B., and E. S. Harris, 1995, The CommodityConsumer Price Connection:
Fact or Fable, Economic Policy Review, October, pp. 2138.
Borio, C., and P. Lowe, 2002, Asset Prices, Financial and Monetary Stability: Exploring
the Nexus, BIS Working Paper 114.
Borio, C., and W. White, 2004, Whither Monetary and Financial Stability? The Implications of Evolving Policy Regimes, in Monetary Policy and Uncertainty: Adapting to a
Changing Economy, pp. 131211 (Federal Reserve Bank of Kansas City).
Browne, F., and D. Cronin, 2010, Commodity Prices, Money and Inflation, Journal of
Economics and Business 62, pp. 33145.
Canuto, O., 2009, The Arrival of Asset Prices in Monetary Policy, RGE EconoMonitor 20,
October, URL: http://www.economonitor.com/blog/2009/10/the-arrival-of-asset-prices
-in-monetary-policy/.
Cody, B. J., and L. D. Mills, 1991, The Role of Commodity Prices in Formulating Monetary
Policy, The Review of Economics and Statistics 73(2), pp. 35865.
275
i
i
i
perrucci 2012/7/11 17:12 page 276 #298
INFLATION-SENSITIVE ASSETS
Cuddington, J. T., and D. Jerrett, 2008, Super Cycles in Real Metals Prices?, IMF Staff
Papers 55(4), pp. 54165.
Dornbusch, R., 1976, Expectations and Exchange Rate Dynamics, Journal of Political
Economy 84, pp. 116176.
Frankel, J., 1984, Commodity Prices and Money: Lessons from International Finance,
American Journal of Agricultural Economics 66(5), pp. 56066.
Frankel, J., 1986, Expectations and Commodity Price Dynamics: The Overshooting
Model, American Journal of Agricultural Economics 68(2), pp. 3448.
Frankel, J., 2008, The Effect of Monetary Policy on Real Commodity Prices, in Campbell,
J. (ed), Asset Prices and Monetary Policy, pp. 29127 (University of Chicago Press).
Furlong, F., and R. Ingenito, 1996, Commodity Prices and Inflation, Economic Review,
Federal Reserve Bank of San Francisco 2, pp. 2747.
Garner, C. A., 1989, Commodity Prices: Policy Target or Information Variable?, Journal
of Money, Credit and Banking 21(4), pp. 50814.
Habermeier, K., I. tker-Robe, L. Jacome, A. Giustiniani, K. Ishi, D. Vvra, T. Kisinbay
and F. Vazquez, 2009, Inflation Pressures and Monetary Policy Options in Emerging and
Developing Countries: A Cross Regional Perspective, IMF Working Paper WP/09/1.
International Monetary Fund, 2005, Does Inflation Targeting Work in Emerging Markets?, World Economic Outlook, September, pp. 16186.
Koo, R., 2011, Commodity Price Increases have Speculative and Structural Roots, URL:
http://www.economist.com/node/21015587.
Marquis, M. H., and S. R. Cunningham, 1990, Is There a Role for Commodity Prices
in the Design of Monetary Policy? Some Empirical Evidence, Southern Economic Journal
57(2), pp. 394412.
Pecchenino, R. A., 1992, Commodity Prices and the CPI: Cointegration, Information and
Signal Extraction, International Journal of Forecasting 7, pp. 493500.
Spiegel, M. M., 2006, Did Quantitative Easing by the Bank of Japan Work , FRBSF
Economic Letter, No. 2006-28, October.
Stark, J., 2007, Objectives and Challenges of Monetary Policy: A View from the ECB,
Speech to Magyar Nemzeti Bank Conference on Inflation Targeting, Budapest, Hungary,
January 19, URL: http://www.ecb.int/press/key/date/2007/html/sp070119.en.html.
Taylor, J. B., 1993, Discretion versus Policy Rules in Practice, Carnegie-Rochester
Conference Series on Public Policy 39, pp. 195214.
Taylor, J. B., 2009, The Financial Crisis and the Policy Responses: An Empirical Analysis
of What Went Wrong. NBER Working Paper 14631.
Ugai, H., 2007, Effects of the Quantitative Easing Policy: A Survey of Empirical Analyses,
Monetary and Economic Studies, March, pp. 148.
Webb, R. H., 1988, Commodity Prices as Predictors of Aggregate Price Change, Economic
Review, Federal Reserve Bank of Richmond, November/December, pp. 311.
276
i
i
i
perrucci 2012/7/11 17:12 page 277 #299
12
John A. Tatom
Indiana State University
i
i
i
perrucci 2012/7/11 17:12 page 278 #300
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 279 #301
i
i
i
perrucci 2012/7/11 17:12 page 280 #302
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 281 #303
i
i
i
perrucci 2012/7/11 17:12 page 282 #304
INFLATION-SENSITIVE ASSETS
PCE
Core PCE
CPI
2009 Q1
2005 Q3
2002 Q1
1998 Q3
1995 Q1
1991 Q3
1988 Q1
1984 Q3
1980 Q1
1977 Q3
1974 Q1
1970 Q3
1967 Q1
1963 Q3
1960 Q1
i
i
i
perrucci 2012/7/11 17:12 page 283 #305
provided evidence that overall PCE deflator inflation is a better predictor of its own future rate than is a core measure. Kiley (2008)
reached the opposite conclusion about how to estimate the underlying trend of inflation. Mankiw et al (2003) provided an earlier survey
on the issues involved in developing inflationary expectations; they
suggest that expectations reflect partial and incomplete updating in
response to news. In Figure 12.1, temporary surges in inflation in
19745, 197981, 1991, 2000 and 20089 reflect sharp increases in the
relative price of energy. The end of the IranIraq war in 1986 and the
IraqKuwait war in 2001 led to declines in the relative price of oil
and energy that were reflected in declines in prices and temporary
drops in inflation.
An important issue that is not explored here is how and whether
asset prices should be included in measures of the price level.
Alchian and Klein (1973) stressed the importance of including assets
in the market basket of expenditures along with goods and services. Others have argued for including the prices of the services of
assets, but not the assets themselves, along with other goods and services. For example, house prices would be excluded, but the owners
equivalent rent or other rental prices of housing would be included.
Stock and Watson (2003) examined evidence on the predictive performance of asset prices for inflation and real output growth, using
data from 38 asset price indicators (mainly asset prices) for seven
member countries of the Organisation for Economic Co-operation
and Development (OECD) for the period 195999. Their review of
the literature and empirical evidence points to the same conclusion,
ie, that some asset prices predict either inflation or output growth in
some countries in some periods. However, no systematic evidence is
found to support the inclusion of asset prices in standard price measures. Bryan et al (2002) found that including asset prices in models
to forecast inflation does have episodic significance in altering the
pattern of inflation forecasts, but that they do not matter much for
economic significance beyond the quality of forecasts from either
overall inflation measures or from core measures of inflation.
A third critical distinction is that inflation can be expected or unexpected. A surprise in inflation will have different effects on asset
prices and economic performance from a rise in expected inflation
that is in line with ex ante expectations. Most surprises in inflation arise from shifts in relative prices, especially food or energy
283
i
i
i
perrucci 2012/7/11 17:12 page 284 #306
INFLATION-SENSITIVE ASSETS
prices, and are temporary. More fundamental determinants of inflation evolve more predictably, and hence the underlying inflation
is also more predictable or expected. The key factor differentiating
expected and unexpected inflation, in terms of the effects on asset
prices, is that capital markets incorporate expected inflation in asset
demand and supply and in pricing of assets, while unexpected inflation is not incorporated in the same way. The key effects of unexpected inflation arise from redistributions of income and wealth. For
example, an unexpected and temporary rise in inflation will redistribute income from creditors, who face an unexpected fall in the
purchasing power of their interest income and principal repayments,
that is, a fall in their realised real rate of return, while debtors enjoy
their corresponding unexpected gain. For owners of firms, there are
at least two effects of unexpected inflation. The first is the redistribution of income from creditors to debtors, as above. Firms that
have external debt will benefit from a rise in unanticipated inflation.
Their owners, stockholders, will realise the unexpected decline in
their real interest payments to their creditors as a gain in real profits.
The discounted value of the gain will accrue to stockholders as an
unanticipated capital gain in the stock price.
Expected inflation usually does not create redistributions of
income or wealth. Market participants require compensation for
expected changes in the purchasing power of expected future payments in order to acquire assets, and sellers of assets have the wherewithal and willingness to provide such compensation, at least in a
simple world where institutions or regulations do not prevent such
compensation. More broadly, it is expected inflation that has more
sweeping effects on asset and other prices and that affects portfolio
allocation, output and economic growth.
IS MONEY A VEIL? THE MONEY NEUTRALITY HYPOTHESIS
The simplest model of inflation and asset prices assumes that there
are no frictions in markets, there are no transaction or adjustment
costs and that information is freely available. In this model, a rise
in expected inflation raises nominal prices for unchanged quantities of goods and services along new and higher expected inflation
paths. It simultaneously raises prices of resources and marginal costs
by equal percentages, at given quantities of goods and services and
resource input flows. Prices move up freely at the specified rates over
284
i
i
i
perrucci 2012/7/11 17:12 page 285 #307
14
12
10
8
% 6
4
2
0
2009 Q2
2005 Q4
2002 Q2
1998 Q4
1995 Q2
1991 Q4
1988 Q2
1984 Q4
1981 Q2
1977 Q4
1974 Q2
1970 Q4
1967 Q2
1963 Q4
1960 Q2
1956 Q4
1953 Q2
i
i
i
perrucci 2012/7/11 17:12 page 286 #308
INFLATION-SENSITIVE ASSETS
interest rate on fixed income securities will rise in line (ie, one to one)
with the expected rate of inflation, while real interest rates and supply and demand for credit will remain unchanged. The relationship
between nominal rates and inflation is clearly shown in Figure 12.2,
where the year-over-year (YoY) PCE inflation is plotted along with
the 10-year Treasury yield.
A rise in expected inflation will cause bond prices to fall, as nominal interest rates rise to incorporate the Fisher effect of inflation on
nominal interest rates. The fall in bond prices is larger, the longer the
duration of the asset, because longer duration cashflows lose more
purchasing power due to a given pace of inflation. The real interest
rate on bonds is unaffected if money growth is neutral.
EXCHANGE RATES AND INFLATION
In this simple model, another key nominal price besides nominal
interest rates is the foreign exchange rate. The benchmark for foreign
exchange in the simple world is determined by relative purchasing
power parity, so that the value of a currency will fall over time relative to another currency, in line with the higher domestic expected
rate of inflation compared with expected inflation abroad. The rate
of depreciation in a local currency will equal the expected inflation
rate in the economy minus the expected inflation rate in the country of origin of the other currency. So long as this holds, the local
currency prices of domestic and foreign goods or services will rise
at the new expected rate of domestic inflation, and the same will be
true abroad.
Even in this simple world, currencies will not adjust fully and
immediately to a change in the expected inflation rate. In fact, currencies typically overshoot in response to factors booting inflation
expectations, falling more than the theory predicts based only on
expected future inflation. This is not unique to exchange rates, as
other asset prices also adjust more in the short run than higher inflation might suggest. The elasticity of supply of some commodities, for
example, is quite inelastic, so that increases in nominal demand lead
to relatively higher relative prices. These relatively large short-run
price adjustments can make commodities an attractive investment
option in the short run. Balanced against this argument, however,
is the short-term nature of these adjustments. Many commodities
are also very sensitive to the business cycle and relatively more
286
i
i
i
perrucci 2012/7/11 17:12 page 287 #309
14
12
6
4
Inflation
10
8
2
2
2009 Q3
2011 Q2
2006 Q1
2002 Q3
1999 Q1
1995 Q3
1992 Q1
1988 Q3
1985 Q1
1981 Q3
1978 Q1
1974 Q3
1971 Q1
1967 Q3
1964 Q1
0
1960 Q3
1,800
1,600
1,400
1,200
1,000
800
600
400
200
0
1957 Q1
S&P index
Source: FRED2, Federal Reserve Bank of St Louis and Standard & Poors.
i
i
i
perrucci 2012/7/11 17:12 page 288 #310
INFLATION-SENSITIVE ASSETS
196584, when the real stock was depressed. Other periods of relatively higher inflation are also associated with relatively low real
stock prices, including the late 1950s and 20079. The decline in inflation in 19918 was notably associated with a sharp rise in the real
stock price. The correlation between the two series is 0.44: strongly
negative.
In fact, the negative correlation between inflation and nominal (or real) stock prices has become one of the most commonly
accepted empirical facts, motivating large numbers of financial and
monetary economics studies. Some of these are briefly mentioned
below, starting with an exception: one study that disagrees with the
empirical finding of the negative relationship between inflation and
stock prices. Konchitchki (2011) suggests that inflation raises both
future real earnings and real stock prices. He provides evidence that
unrecognised accounting inflation gains raise future cashflows, and
yield abnormally high returns on equities, reflecting a failure of the
market to fully account for future gains in nominal cashflow when
inflation initially occurs.
The most widely accepted hypothesis of a negative effect of inflation on stock prices is that supply shocks, which reduce output and
raise prices, also reduce the marginal productivity of capital, reducing the real earnings of capital and lowering the value of capital
assets and the firm.
Bakshi and Chen (1996) suggested that the negative relation of
inflation and stock prices occurs because of procyclical movements
in real interest rates and in inflation. A cyclical expansion (decline),
in their view, will raise the real rate of interest and also raise inflation. The higher real interest rate would lower stock prices, giving
rise to the negative correlation of stock prices and inflation. However, the empirical support is rather weak, not surprisingly, because
it depends on controversial empirical assertions of cyclical real interest rates and the existence of a Phillips (1958) curve. Fama (1981) and
Stulz (1986) relied on another channel: a rise in expected inflation
reduces wealth and this, in turn, lowers the expected return on equities and lowers stock prices. Fama (1981) also attributed the negative
relationship to supply shocks, again operating via a wealth effect.
Brandt and Wang (2003) provided an alternative explanation of
the negative relationship of nominal stock prices and unexpected
inflation. They argued that unexpected inflation affects investor risk
288
i
i
i
perrucci 2012/7/11 17:12 page 289 #311
i
i
i
perrucci 2012/7/11 17:12 page 290 #312
INFLATION-SENSITIVE ASSETS
t
1t
d ,
dn = dr + d =
t
1t
i
i
i
perrucci 2012/7/11 17:12 page 291 #313
i
i
i
perrucci 2012/7/11 17:12 page 292 #314
INFLATION-SENSITIVE ASSETS
2010 Q1
2005 Q3
2001 Q1
1996 Q3
1992 Q1
1987 Q3
1983 Q1
1978 Q3
1974 Q1
1969 Q3
1965 Q1
1960 Q3
1956 Q1
Economic profit
Measured profit
1951 Q3
12
11
10
9
8
7
6
5
4
3
2
1947 Q1
i
i
i
perrucci 2012/7/11 17:12 page 293 #315
(7.11)
adjusted R = 0.93,
(3.20)
(2.23)
SE = 0.051,
(43.97 )
DW = 2.07
The standard error (SE) of the estimate and the DurbinWatson (DW)
statistic are both satisfactory. A one-percentage point rise in inflation
reduces economic profits relative to measured profits by 1.8%, as we
would expect. However, a rise in the capacity utilisation rate (our
business-cycle variable) also acts in the same direction, reducing
economic profit relative to measured profits: a less intuitive result,
which will be explained in the following.
The results from this regression support the hypothesis that inflation reduces the difference between economic and measured profits,
but it does not reveal how inflation affects each individual measure
individually. To this end, consider two separate regressions, one linking after-tax economic profits to inflation and the business cycle and
293
i
i
i
perrucci 2012/7/11 17:12 page 294 #316
INFLATION-SENSITIVE ASSETS
SE = 0.062,
DW = 1.88
SE = 0.070,
DW = 2.30
i
i
i
perrucci 2012/7/11 17:12 page 295 #317
1.0
50
0.5
40
60
2005
1.5
1999
2.0
70
1993
80
1984
2.5
1978
3.0
90
1972
3.5
100
1966
110
1960
4.0
1954
120
1948
Equipment vs structures
(%, 2005 = 100)
i
i
i
perrucci 2012/7/11 17:12 page 296 #318
INFLATION-SENSITIVE ASSETS
REFERENCES
Alchian, A. A., and B. Klein, 1973, On a Correct Measure of Inflation, Journal of Money,
Credit, and Banking 5(1), pp. 17391.
Bakshi, G. S., and Z. Chen, 1996, Inflation, Asset Prices, and the Term Structure of Interest
Rates in Monetary Economies, Review of Financial Studies 9, pp. 24176.
Barro, R. J., 1996, Inflation and Growth, Federal Reserve Bank of St Louis Review 48(3),
pp. 15369.
296
i
i
i
perrucci 2012/7/11 17:12 page 297 #319
Bernanke, B. S., 2004, The Great Moderation, Speech to Eastern Economics Association, February 20, URL: http://www.federalreserve.gov/boarddocs/speeches/2004/
20040220/default.htm.
Blanchard, O., and J. Simon, 2001, The Long and Large Decline in US Output Volatility,
Brookings Papers on Economic Activity 1, pp. 13564.
Bordo, M. D., and D. C. Wheelock, 2007, Stock Market Booms and Monetary Policy in
the Twentieth Century, Federal Reserve Bank of St Louis Review 89(2), pp. 91122.
Brandt, M. W., and K. Q. Wang, 2003, Time Varying Risk Aversion and Unexpected
Inflation, Journal of Monetary Economics 50, pp. 145798.
Bryan, M. F., S. G. Cecchetti, and R. OSullivan, 2002, Asset Prices in the Measurement
of Inflation, NBER Working Paper 8700.
Crone, T., N. Neil, K. Khettry, L. Mester and J. Novak, 2011, Core Measures of Inflation as
Predictors of Total Inflation, Federal Reserve Bank of Philadelphia Working Paper 11-24.
Fama, E. F., 1981, Stock Returns, Real Activity, Inflation, and Money, American Economic
Review 71, pp. 54565.
Fama, E. F., and K. R. French, 1989, Stock Returns, Expected Returns, and Real Activity,
Journal of Financial Economics 45, pp. 10891108.
Feldstein, M., 1981, Inflation and the Stock Market, American Economic Review, September, pp. 83947.
Fisher, I., 1907, The Rate of Interest (New York: Macmillan).
Fisher, I., 1911, The Purchasing Power of Money (New York: Macmillan).
Friedman, M., 1956, The Quantity Theory of Money: A Restatement, in M. Friedman
(ed), Studies in the Quantity Theory of Money, pp. 321 (University of Chicago Press).
Hess, P. J., and B.-S. Lee, 1999, Stock Returns and Inflation with Supply and Demand
Disturbances, Review of Financial Studies 12(5), pp. 120318.
Kiley, M. T., 2008, Estimating the Common Trend Rate of Inflation for Consumer Prices
and Consumer Prices excluding Food and Energy Prices, Board of Governors of the
Federal Reserve System Finance and Economic Discussion Series 2008-38, August.
Kim, C.-J., and C. Nelson, 1999, Has the US Economy Become More Stable? A Bayesian
Approach Based on a Markov-Switching Model of the Business Cycle, Review of Economics
and Statistics 81, pp. 60816.
Kim, C.-J., C. Nelson and J. Piger, 2004, The Less Volatile US Economy: A Bayesian Investigation of Timing, Breadth, and Potential Explanations, Journal of Business and Economic
Statistics 22(1), pp. 8093.
Konchitchki, Y., 2011, Inflation and Nominal Financial Reporting: Implications for
Performance and Stock Prices, The Accounting Review 86(3), pp. 104585.
Mankiw, N. G., R. Reis and J. Wolfers, 2003, Disagreements about Inflation Expectations, in NBER Macroeconomics Annual, pp. 20970 (Boston, MA: MIT Press).
McConnell, M., and G. Prez-Quirs, 2000, Output Fluctuations in the United States:
What Has Changed since the Early 1980s?, American Economic Review 90, pp. 146476.
Patinkin, D., 1965, Money, Interest and Prices: An Integration of Monetary and Value Theory,
Second Edition (New York: Harper and Row).
Phillips, A. W., 1958, The Relationship between Unemployment and the Rate of Change
of Money Wages in the United Kingdom 18611957, Economica 25(100), pp. 28399.
297
i
i
i
perrucci 2012/7/11 17:12 page 298 #320
INFLATION-SENSITIVE ASSETS
Piazzesi, M., and M. Schneider, 2008, Inflation and the Price of Real Assets, Stanford
University Working Paper, April.
Rasche, R. H., and J. A. Tatom, 1977, The Effect of the New Energy Regime on Economic
Capacity, Production, and Prices, Federal Reserve Bank of St Louis Review, May.
Rasche, R. H., and J. A. Tatom, 1981, Energy Price Shocks, Aggregate Supply and Monetary Policy: The Theory and International Evidence, in K. Brunner and A. H. Meltzer
(eds), Supply Shocks, Incentives and National Wealth, Carnegie-Rochester Conference Series
on Public Policy 14, pp. 993.
Rich, R. W., and C. Steindel, 2007, A Comparison of Measures of Core Inflation, Federal
Reserve Bank of New York Economic Policy Review 13(3), pp. 1938.
Stock, J. H., and M. W. Watson, 2003, Forecasting Output and Inflation: The Role of Asset
Prices, Journal of Economic Literature 41(3), pp. 788829.
Stulz, R. M., 1986, Asset Pricing and Expected Inflation, The Journal of Finance 41(1),
pp. 20923.
Tatom, J. A., 2002, Stock Prices, Inflation and Monetary Policy, Business Economics,
October, pp. 719.
Tatom, J. A., and J. E. Turley, 1978, Inflation and Taxes: Disincentives for Capital Formation, Federal Reserve Bank of St Louis Review, January. Reprinted in Federal Reserve Readings
on Inflation, Federal Reserve Bank of New York, February 1979, pp. 16773.
298
i
i
i
perrucci 2012/7/11 17:12 page 299 #321
13
Jeffrey Oxman
University of St Thomas
i
i
i
perrucci 2012/7/11 17:12 page 300 #322
INFLATION-SENSITIVE ASSETS
Nominal
Real
Inflation
Nominal
Real
Inflation
Nominal
Real
Inflation
Nominal
Real
Inflation
Nominal
Real
Inflation
Brazil
Canada
Chile
Mexico
US
17.30
7.84
8.77
7.25
5.33
1.82
9.02
5.05
3.78
17.42
7.22
9.51
6.30
3.88
2.32
Austria
Denmark
France
5.90
4.13
1.70
9.78
7.68
1.96
4.40
2.86
1.50
7.43
5.96
1.38
Ireland
Israel
Italy
Norway
Spain
4.72
2.25
2.41
12.21
8.36
3.55
1.37
3.40
2.11
9.68
7.53
2.00
7.26
4.44
2.69
Switzerland
UK
Australia
China
India
5.60
4.72
0.84
4.09
2.14
1.91
5.85
3.13
2.64
13.86
10.99
2.58
10.91
4.21
6.43
Japan
Korea
New Zealand
4.10
4.03
0.08
4.32
0.87
3.42
Germany Greece
3.04
1.00
4.09
0.75
1.56
2.35
Nominal returns data are obtained from Bloomberg for the broadest stock
index with the longest history for each country. Inflation is the Consumer
Price Index (CPI) as measured by the Organization for Economic Cooperation and Development (OECD). Inflation data were obtained from
the Federal Reserve Bank of St Louis.
i
i
i
perrucci 2012/7/11 17:12 page 301 #323
30
(a)
30
(b)
20
% 20
10
10
0
0
0.5
30
0.5
0.5
40
(c)
0.5
1.0
(d)
30
20
20
%
10
0
1.0
10
0
0.5
0.5
1.0
1.0 0.5
30
25 (e)
0.5
1.0
1.5
(f)
20
%
20
15
10
10
5
0
1.0
0.5
0
0.5
Returns
1.0
Returns
(a) US, (b) Japan, (c) Germany, (d) India, (e) Mexico, (f) Brazil. Solid line, nominal (estimated); dashed line, real (estimated). Grey histogram, nominal; black
histogram, real.
i
i
i
perrucci 2012/7/11 17:12 page 302 #324
INFLATION-SENSITIVE ASSETS
40
20
30
15
% 20
10
10
0
5
0.02
0
0.02 0.04
Annual inflation
0.06
0.20
80
200
60
150
% 40
100
20
0
50
0
0.5
1.0
Annual inflation
1.5
0.2
0.4
0.6
Annual inflation
0.8
Inflation in: (a) US, Germany and Japan (solid line, US; dashed line, Germany;
dotted line, Japan; light-grey histogram, US; dark-grey histogram, Germany; black
histogram, Japan); (b) India; (c) Mexico; (d) Brazil.
i
i
i
perrucci 2012/7/11 17:12 page 303 #325
(13.2)
(13.3)
Therefore, assuming that risk premium does depend on expected inflation, stock returns should also increase one-for-one with
expected inflation
rN
rstock
=
=1
(13.4)
iEXP
iEXP
Furthermore, if we assume that itself is not dependent on expected
inflation, the key empirical issue is the relationship between the
equity risk premium and inflation, or equivalently between inflation
and the overall market return rMkt .
Although many of the simplifying assumptions above have been
challenged by the empirical data, Fishers hypothesis provides a
simple yet powerful starting point in the analysis of the relationship
between stock returns and inflation. Specifically, this equation predicts nominal stock returns to be positively correlated with ex ante
303
i
i
i
perrucci 2012/7/11 17:12 page 304 #326
INFLATION-SENSITIVE ASSETS
regimes).
EXPECTED VERSUS UNEXPECTED INFLATION
Expected inflation is the percentage price increase that investors
think will take place over a certain time period in the future, based
on information available in the present. By definition, only expected
inflation can be incorporated into the required return calculation for
an asset. However, unexpected inflation, or the difference between
expected and actual inflation, has potential consequences for stock
returns.
Among its effects, unexpected inflation constitutes a wealth transfer from net creditors to net debtors, increases the real tax burden
of firms with significant fixed assets and inventory and it indicates
that future inflation may be different from previous expectations. If
unexpected inflation is positive, then future expected inflation will
increase and cause an increase in required returns.
The transfer of wealth from creditors to debtors results because
contracts are written in US dollar terms. This transfer will exist for
all dollar-denominated contracts that do not adjust for unexpected
inflation, or are renegotiated infrequently. Thus, unexpected inflation may also constitute a wealth transfer from the labour force to the
shareholders because wages will typically not rise as fast as inflation,
so profit margins should improve at least over the short term. The
stock-return effect on any specific firm is a combination of several
effects and thus unclear a priori.
In addition to the automatic effects illustrated above, inflation
surprises might also lead to government intervention to counteract
high inflation or avoid deflation. Examples include Nixon-era price
304
i
i
i
perrucci 2012/7/11 17:12 page 305 #327
i
i
i
perrucci 2012/7/11 17:12 page 306 #328
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 307 #329
Because Famas model does not account for all the negative correlation between unexpected inflation and stock returns, Geske and
Roll (1983) proposed an extension to Famas theory. Their argument
focuses on the effect of stock market returns on government finances.
Essentially, government revenues are comprised of personal and corporate taxes, which respond to expected economic conditions in the
same direction as the stock market. Thus, stock market returns anticipate changes in government revenue. Government expenditure, on
the other hand, is largely fixed. Clearly, government deficits are
countercyclical (they tend to decrease when the economy is strong,
and increase when the economy is weak). Geske and Roll hypothesise that a decrease in the stock market will lead to higher deficits
and, given that under such circumstances the Federal Reserve System will have an incentive to monetise the growing debt, this in turn
will lead to growth in-the-money supply and higher inflation. Thus,
the measured correlation between stock returns and inflation is negative. This hypothesis is also called the reverse-causality link, as
it is declining stock prices that cause inflation to rise, and not the
other way around.
Regarding point 2 above, Danthine and Donaldson (1986) developed a general equilibrium model and showed that common stocks
are not a good hedge against non-monetary inflation, ie, a price level
increase caused by a negative supply shock (for example, oil prices in
the 1970s); however, common stocks are a good hedge against monetary inflation (typically associated with a positive demand shock2 )
over the long run. This study differentiates between inflation caused
by supply shocks and inflation caused by demand shocks (the latter being more equity friendly than the former) and recognises the
importance of the horizon used in the analysis, as different dynamics
might be at play in the short and the long term.
Marshall (1992) developed and estimated a representative agent
model where agents hold a mix of money, equities and bonds as
assets. The value of money is that it reduces the costs of consumption
transactions. Agents seek to maximise the present value of expected
consumption utility. The agents do not have income; rather, they are
constrained by the amount of shares and bonds they hold. Agents
can lend and borrow (issue bonds) at the risk-free real rate of interest. Finally, increases in money supply are exogenous (the money
supply function follows a stochastic process) and new money is
307
i
i
i
perrucci 2012/7/11 17:12 page 308 #330
INFLATION-SENSITIVE ASSETS
distributed directly to agents as a lump-sum transfer. The importance of Marshalls model is that it does not include any irrationality or market inefficiencies, which other authors have suggested as
explanations of the negative correlation between real equity returns
and inflation (see the discussion in the next section regarding inflation illusion). It is clear that the Fisher hypothesis is violated by
the negative correlation between expected inflation and expected
stock returns. Marshalls contribution shows that this violation of
the Fisher hypothesis arises because of the importance of money
for transactions. Marshalls model also shows that the hypothesis of
Fama (1981) is correct, in that it is fluctuations in the real economy
that cause equity returns and inflation to move in opposite directions. Finally, an increase in expected inflation that arises due to an
increase in expected money growth causes agents to hold less money
and therefore increase balances held in equities or bonds. Marshalls
model suggests there should be limited or no effects from money
supply-induced inflation and equity returns.3
While the theoretical work reviewed to this point does not rely
on financial intermediaries, the work of Boyd et al (2001) points to
the banking sector as an important agent for transmitting inflationinduced problems to the economy as a whole, including equity
prices. In addition to new insights, they have one of the largest
data samples of any papers reviewed: 100 countries from 19601995.
Inflation affects the real activity through the banking sector because
higher inflation, even if predictable, exacerbates information asymmetries that already exist in credit markets. As inflation increases, the
informational frictions become more important and result in credit
rationing. With fewer loans being made, fewer projects can be started
and thus real output growth declines.
Boyd et al (2001) showed that countries with high inflation (their
cut-off is 15% per year) have less developed financial sectors than
countries with low inflation. This leads to some differences in the
effects of inflation on equity returns. In low-to-moderate inflation
countries, inflation rates and stock market liquidity and trading volume are inversely correlated, while inflation is positively correlated
with equity return volatility. But, above 15% inflation per annum,
the correlations between inflation and equity market activity disappear. Finally, below 15% inflation they observed no correlation
between nominal equity returns and inflation, but above 15% they
308
i
i
i
perrucci 2012/7/11 17:12 page 309 #331
found a one-to-one positive correlation between inflation and nominal equity returns. Thus, when examining foreign countries for
potential investment, keeping in mind the 15% inflation cut-off is
likely to be helpful.
Stock price based theories of inflation and asset returns
It is helpful to review the Gordon (1962) growth model of stock
pricing, which states that the value of a share of equity today is the
present value of the dividends accruing to the share in perpetuity.
Formally
Dt+1
(13.5)
Pt = stock
r
g
where Pt the current price of one share, Dt+1 the expected dividend
next period, rstock is the required return on the stock and g is the
expected growth rate of the dividend. The Gordon model implies
that stock prices should not be affected by expected inflation, as
the inflation contribution cancels in the difference rstock g. However, we know that empirical data does not support this conclusion.
The previous section covered various economic reasons for this phenomenon. This section explores reasons specifically related to stock
pricing models.
There are three possible reasons for stock prices to be affected
by inflation, according to the Gordon model. The first is that the
Fisher hypothesis for dividend growth rates is incorrect, and real
dividend growth is actually a function of expected inflation, among
other things. Second, the Fisher hypothesis for stock returns is incorrect, and the required real stock return is affected by expected inflation. Finally, there is the inflation illusion hypothesis, developed
by Modigliani and Cohn (1979).
The inflation illusion hypothesis states that investors mistakenly
extrapolate past nominal dividend growth into the future, even
when inflation is changing. For example, if inflation is expected to
increase, investors operating under the inflation illusion hypothesis
will increase their estimate of nominal required return rstock but will
not adjust their estimate of nominal growth. This leads to lower stock
prices, and a negative correlation between inflation and stock prices.
The reverse is also the case: when expected inflation decreases, stocks
become overvalued.
Sharpe (2002) provided the first set of results in our overview. He
found that the negative relationship between stock valuation and
309
i
i
i
perrucci 2012/7/11 17:12 page 310 #332
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 311 #333
Note that the independent variables on the left represent realised historical data (for example, the realised dividend growth rate gex post ),
while the dividend price ratio on the left embeds (by definition)
future expectations of the same quantity, ie, gex ante . The motivation
for the regression is that if the dividend growth rate and subjective
risk premium explain the dividend/price ratio, then the hypothesis
of inflation illusion is rejected. If the error term is significant and
positively related to inflation, then the hypothesis of inflation illusion cannot be rejected. The results from the analysis of Campbell
and Vuolteenaho (2004) indicate that the realised dividend growth
rate gex post and inflation are positively related. Thus, empirically
speaking, inflation does not have an adverse impact on the realised
real growth of dividends. Also, the (subjective) risk premium is not
related to inflation, so investors do not become more risk averse
when inflation increases. The mispricing component is strongly positively related to inflation, and this constitutes evidence in favour of
the inflation illusion hypothesis (ie, that the ex ante projection for
nominal dividend growth does not fully account for inflation, so
that r stock gex ante is a positive relation to inflation). Campbell and
Vuolteenaho also suggested that, while inflation illusion is present
in the short run, it should diminish over the long run (longer than
one year); therefore, equities tend to be a poor inflation hedge in the
short run, but a better hedge over the long run. This is consistent
with results that we shall review in the next section.
The previous sets of results involve expected inflation. But it might
be the case that the inflationequity return correlation is caused by
unexpected inflation. Sudden price changes in a commodity or service might occur because of a sudden and unexpected change in
production. Negative supply shocks are those that cause supply to
decrease and thus the price of the commodity to increase. Positive
supply shocks are the opposite, and are often termed technology
shocks because of the close positive correlation between technological innovation and productive capacity. If the commodity in question is particularly ubiquitous in the economy, like crude oil, this
price shock can work its way through the economy generally, thus
showing up as inflation.
Demand shocks are sudden changes in the demand for a certain
commodity or service. Of course, demand shocks have the opposite
effect to supply shocks. A particularly important type of demand
311
i
i
i
perrucci 2012/7/11 17:12 page 312 #334
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 313 #335
i
i
i
perrucci 2012/7/11 17:12 page 314 #336
INFLATION-SENSITIVE ASSETS
monetary policy, allowing money supply to rise and fall with gross
national product, then stock returns and inflation would be positively related. Kaul (1987) shows this is the case for the depression
period in the US and Canada.
In a follow up paper in 1990, Kaul considered the impact of two
different types of monetary policy regimes on the stock-return
inflation correlation. There are two ways that central banks carry
out their mandate of price stability and, in the case of the US Federal
Reserve (the Fed), the additional mandate of full employment. The
central bank can target a specific growth rate in the monetary base, or
the central bank can target interest rates, like the discount window
rate or the overnight federal funds rate. The monetary base is the
total currency held by the public plus vault cash and deposits held
by Federal Reserve banks. The discount window rate is the interest
rate charged by the Fed on overnight lending directly from the Fed
to its member banks. The federal funds rate is the rate at which banks
lend to each other in the overnight lending market.
Kaul (1990) identified periods in which central banks in four countries (the US, Canada, the UK, West Germany) follow the two different methods discussed above. When monetary policy targets money
growth directly, there is no clear link between money supply and economic activity, and thus there should be no significant correlation
between inflation and equity returns.
Kaul followed Geske and Roll (1983) in assuming that if central banks use interest rate targeting as the mechanism of implementing policy, then the central banks will follow a countercyclical policy. This is because interest rate targeting leads the Fed to
monetise government debt, giving rise to the mechanism identified
by Geske and Roll (1983) and leading to the negative correlation
between stock returns and inflation. Kauls results confirm that the
reason for the apparent shift from no relation or a positive relation
between inflation and equity returns to a negative relation is due to
the shift in how the central banks implement their mandates. In the
US and Canada, central banks moved from targeting money supply to focusing on interest rates, especially after the oil crisis in the
1970s. Moreover, Kaul highlighted that, unlike previous work, it is
the interaction of money demand and money supply that governs
the relationship between inflation and equity returns, not just money
demand.
314
i
i
i
perrucci 2012/7/11 17:12 page 315 #337
The message of Kaul (1987, 1990) is that in industrialised countries where the central bank uses interest rate management to fulfill its objective of price stability (and full employment as in the
US), investors should expect a negative relationship between stock
returns and inflation, both expected and unexpected. Therefore,
stocks will not be a satisfactory inflation hedge in the short run.
Chang and Pinegar (1987) investigated the effect of security risk
on the relationship between stock returns and inflation. They found
that the riskier the security, as measured by the market beta, the
more negative the correlation between inflation (expected and unexpected) and real stock returns. However, their results are based on
the model developed by Geske and Roll (1983), which posited a monetary growth mechanism rather than an interest rate management
mechanism. Thus, it is not clear if Chang and Pinegars results are
applicable to the environment of the 2000s.
In the spirit of Danthine and Donaldson (1986), Boudoukh and
Richardson (1993) offered the first exploration of the stock-return
inflation relationship over a long horizon. They used stock returns
from 1802 to 1990 and a five-year time horizon. Previous studies
used post-World War II data and month/quarter/year horizons.
For the longer time horizon, Boudoukh and Richardson (1993)
found that stock returns do compensate for inflation, but not fully.
A 1% increase in inflation yields a 0.5% increase in nominal stock
returns over a five-year period. The one-year horizon conforms to
previous studies, where inflation-hedging performance is generally
poor. Thus, over the long run, stocks offer some inflation protection.
They confirmed this result for the UK over the period 18201988.
Gregoriou and Kontonikas (2010) updated the work of Boudoukh
and Richardson (1993) with a shorter (19702006) but wider (16
OECD countries) sample. Their findings indicate that over the long
run (ie, 10 to 16 years), annual elasticity of stock prices to consumer
goods prices is not significantly different from unity. That means
that, over the long run, stocks will provide a nearly one-to-one hedge
against inflation. We have found that these results are the strongest
in favour of stocks as an inflation hedge.
Schotman and Schweitzer (2000) derived optimal hedge portfolios for various assumptions regarding inflation persistence and
the long-run relationship between equity returns and inflation. Their
results indicate that only when inflation is fairly persistent and stock
315
i
i
i
perrucci 2012/7/11 17:12 page 316 #338
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 317 #339
i
i
i
perrucci 2012/7/11 17:12 page 318 #340
INFLATION-SENSITIVE ASSETS
firms with large pension expenses (if the latter are fixed in nominal terms). Depreciation does not appear to have any significant
effect. Thus, to summarise, if we want a better inflation hedge, we
should avoid firms that carry large inventories, and seek to hold net
debtors.
Boudoukh et al (1994) investigated the negative correlation by disaggregating the stock market into industries. They found that, in the
short run, non-cyclical industries tend to have a positive correlation
with expected inflation, while cyclical industries have a negative
correlation. Over the entire sample in their paper (19531990), most
industries have a positive correlation. It appears that those industries
that are related to consumer goods have a much stronger relationship
to consumer inflation than industries that are related to producers
goods. Thus, if we are looking for an inflation hedge over the long
run, it is best to focus on industries such as tobacco, apparel or food
and beverage.
Ely and Robinson (1997) updated the literature by taking an international view, offering evidence in favour of stocks as a hedge
against inflation. Their work indicates that focusing on the US may
deliver anomalous results. In the US, the efficacy of stocks as a hedge
against inflation depends on the source of inflation. If inflation occurs
because of a real output shock, then stocks do not offer a good hedge.
If it is monetary-induced inflation, then stocks do offer a good hedge.
Their work is based on a vector error-correction model (VECM) and
uses a 16-quarter horizon; it also covers Canada, Western Europe,
Australia, the UK and Japan. Many of the countries display their
own idiosyncratic relationship to inflation; thus, it is valuable for
the reader to investigate the specific country of interest. Some important observations are in order though. Canada and Italy both have
similar responses to the US to output shocks. Switzerland does not
offer protection from inflation, as stock prices decrease for both real
output and monetary shocks. Spain is the only country to offer solid
protection from output-related inflation.
Brire and Signori (2009) make an important contribution to the
asset allocation literature, but did not focus exclusively on equities.
Working in a real-return maximising portfolio environment, they
found that equity weight should be increased in a portfolio when
the targeted real return is greater than zero or the investment horizon
is long (typically two years or more).
318
i
i
i
perrucci 2012/7/11 17:12 page 319 #341
i
i
i
perrucci 2012/7/11 17:12 page 320 #342
INFLATION-SENSITIVE ASSETS
returns in Chile, Israel, Mexico and Peru all have high positive correlations with inflation, and these are the four highest inflation countries in the sample. It appears that in high inflation countries there
is sufficient compensation through nominal equity returns to enjoy
a sound inflation hedge, as nominal equity returns adjust close to
one-to-one with inflation.
Barnes et al also explored the effects of spillover inflation from
the US into other countries. They found that for Germany,8 Luxembourg, the Netherlands, Peru, Portugal, Switzerland and the UK
there is a significant negative relationship between equity returns in
the country and US inflation. Only in the case of Israel is the spillover
positive. As Amihud (1996) showed, Israel is a special case due to
the differing institutional treatment of inflation.
Choudhry (2001) examined four countries that experienced high
inflation from the 1980s to the end of the 1990s: Argentina, Chile,
Mexico and Venezuela. Nominal stock returns were, on average,
higher than inflation. The efficacy of stocks as an inflation hedge
depends on the country under investigation. Argentina and Chile
offer the strongest inflation hedge, with Chile having a slightly larger
compensation than Argentina. Both are close to a one-to-one relationship between nominal returns and contemporaneous inflation.
Mexico also offers some inflation protection, but it is lagged one
month compared with Argentina and Chile. In Venezuela, however,
no inflation protection can be expected.
Choudhrys results are somewhat puzzling, in that Argentina has
the highest inflation in the group, and Chile has the lowest. This
may be because Choudhry does not calculate the source of the inflation: real output shocks or monetary policy. As we discussed earlier,
knowledge of the root cause of inflation is of utmost importance.
CONCLUSION
At this point investors may be quite discouraged regarding the inflation-hedging ability of equities: the bulk of the evidence does not
favour equities in this role. Nevertheless, there are many lessons to
be found in the above discussion.
First, in order to be a good hedge against inflation, equities should
be held over the long term. The Fisher relationship is more likely to
hold for investment horizons of five years or longer than for shortterm investment horizons. Thus, selling after a large inflation shock
320
i
i
i
perrucci 2012/7/11 17:12 page 321 #343
that causes equity prices to decrease is likely to be counterproductive. A better response would be to hold on and wait for equity prices
to restore themselves.
Second, equities in high-inflation regimes offer better inflation
protection than equities in low-inflation regimes. Latin America is
the test case for this in most of the literature. Even in the short run,
countries like Brazil and Argentina offer an inflation hedge much
closer to one-to-one than more developed economies such as the
US and the UK. Of course, we are trading off against other risks by
investing in emerging markets.
Third, there is some evidence that defensive industries, like
tobacco and food and beverage, provide better inflation hedges than
more cyclical industries. This evidence is not strong, however, and
out-of-sample testing suggests past inflation betas are not reliable
guides to future inflation betas.
Equities may not be a reliable inflation hedge in a low-inflation
environment. They do, however, offer a positive real return in most
countries, which is some protection against inflation.
1
In other words, easy monetary conditions (low real interest rates) increase the demand for
money and possibly consumption.
See Bakshi and Chen (1996) for an analysis of a broader array of model economies. They find
that conclusions similar to Marshalls hold for a variety of assumptions about utility functions
and money supply dynamics.
Evidence was provided for West Germany and reunified Germany after 1989.
Total return on stocks is measured as the value-weighted portfolio of all stocks included in
the Center for Research in Securities Prices (CRSP) database. This includes all stocks on the
NYSE, AMEX and Nasdaq exchanges.
Adams et al (2004) use a variety of time horizons, including calendar time (eg, 15-minute
intervals) and transaction time (tick data).
REFERENCES
Adams, G., G. McQueen and R. Wood, 2004, The Effects of Inflation News on High
Frequency Stock Returns, The Journal of Business 77, pp. 54774.
Amihud, Y., 1996, Unexpected Inflation and Stock Returns Revisited: Evidence from
Israel, Journal of Money, Credit and Banking 28, pp. 2233.
Ang, A., M. Brire and O Signori, 2011, Inflation and Individual Equities, Working
Paper, URL: http://ssrn.com/abstract=1805525.
321
i
i
i
perrucci 2012/7/11 17:12 page 322 #344
INFLATION-SENSITIVE ASSETS
Bakshi, G. S., and Z. Chen, 1996, Inflation, Asset Prices, and the Term Structure of Interest
Rates in Monetary Economies, The Review of Financial Studies 9, pp. 24175.
Barnes, M., J. H. Boyd and B. D. Smith, 1999, Inflation and Asset Returns, European
Economic Review 43, pp. 73754.
Bekaert, G., and X. Wang, 2010, Inflation Risk and the Inflation Risk Premium, Working
Paper, URL: http://ssrn.com/abstract=1600312.
Boudoukh, J., and M. Richardson, 1993, Stock Returns and Inflation: A Long-Horizon
Perspective, The American Economic Review 83, pp. 134655.
Boudoukh, J., M. Richardson and R. F. Whitelaw, 1994, Industry Returns and the Fisher
Effect, The Journal of Finance 49, pp. 15951615.
Boyd, J. H., R. Levine and B. D. Smith, 2001, The Impact of Inflation on Financial Sector
Performance, Journal of Monetary Economics 47, pp. 22148.
Brire, M., and O. Signori, 2009, Inflation-Hedging Portfolios in Different Regimes,
Working Paper, URL: http://www.institutlouisbachelier.org/risk10/work/4302987.pdf.
Campbell, J. Y., and R. J. Shiller, 1988, The Dividend-Price Ratio and Expectations of
Future Dividends and Discount Factors, The Review of Financial Studies 1, pp. 195228.
Campbell, J. Y., and T. Vuolteenaho, 2004, Inflation Illusion and Stock Prices, The
American Economic Review 94, pp. 1923.
Chang, E. C., and J. M. Pinegar, 1987, Risk and Inflation, The Journal of Financial and
Quantitative Analysis 22, pp. 8999.
Choudhry, T., 2001, Inflation and Rates of Return on Stocks: Evidence from High Inflation
Countries, Journal of International Financial Markets, Institutions, and Money 11, pp. 7596.
Danthine, J.-P., and J. B. Donaldson, 1986, Inflation and Asset Prices in an Exchange
Economy, Econometrica 54, pp. 585605.
Ely, D. P., and K. J. Robinson, 1997, Are Stocks a Hedge against Inflation? International
Evidence Using a Long-Run Approach, Journal of International Money and Finance 16,
pp. 14167.
Fama, E. F., 1981, Stock Returns, Real Activity, Inflation, and Money, The American
Economic Review 71, pp. 54565.
Fisher, I., 1930, The Theory of Interest (New York, MacMillan).
Geske, R., and R. Roll, 1983, The Fiscal and Monetary Linkage Between Stock Returns
and Inflation, The Journal of Finance 38, pp. 133.
Gordon, M., 1962, The Investment, Financing, and Valuation of the Corporation (Homewood,
IL: Irwin).
Goto, S., and R. Valkanov, 2002, The Feds Effect on Excess Returns and Inflation Is
Bigger than You Think, Working Paper, URL: http://www.personal.anderson.ucla.edu/
rossen.valkanov/page1.htm.
Gregoriou, A., and A. Kontonikas, 2010, The Long-run Relationship Between Stock
Prices and Goods Prices: New Evidence from Panel Cointegration, Journal of International
Financial Markets, Institutions, and Money 20, pp. 16676.
Hess, P. J., and B.-S. Lee, 1999, Stock Returns and Inflation with Supply and Demand
Disturbances, The Review of Financial Studies 12, pp. 120318.
Kaul, G., 1987, Stock Returns and Inflation: The Role of the Monetary Sector, Journal of
Financial Economics 18, pp. 25376.
322
i
i
i
perrucci 2012/7/11 17:12 page 323 #345
Kaul, G., 1990, Monetary Regimes and the Relation between Stock Returns and
Inflationary Expectations, The Journal of Financial and Quantitative Analysis 25, pp. 30721.
Knif, J., J. Kolari and S. Pynnnen, 2008, Stock Market Reaction to Good and Bad
Inflation News, The Journal of Financial Research 31, pp. 14166.
Lee, B.-S., 2010, Stock Returns and Inflation Revisited: An Evaluation of the Inflation
Illusion Hypothesis, Journal of Banking and Finance 34, pp. 125773.
Lintner, J., 1965, The Valuation of Risk Assets and the Selection of Risky Investments in
Stock Portfolios and Capital Budgets. Review of Economics and Statistics 47, pp. 1337.
Marshall, D. A., 1992, Inflation and Asset Returns in a Monetary Economy, The Journal
of Finance 47, pp. 131542.
Modigliani, F., and R. A. Cohn, 1979, Inflation, Rational Valuation, and the Market,
Financial Analysts Journal 35, pp. 2444.
Pearce, D. K., and V. V. Roley, 1988, Firm Characteristics, Unanticipated Inflation, and
Stock Returns, The Journal of Finance 43, pp. 96581.
Polk, C., S. Thompson and T. Vuolteenaho, 2006, Cross-Sectional Forecasts of the Equity
Premium, Journal of Financial Economics 81, pp. 10141.
Schotman, P. C., and M. Schweitzer, 2000, Horizon Sensitivity of the Inflation Hedge of
Stocks, Journal of Empirical Finance 7, pp. 30115.
Schwert, G. W., 1981, The Adjustment of Stock Prices to Information about Inflation,
The Journal of Finance 36, pp. 1529.
Sharpe, S. A., 2002, Reexamining Stock Valuation and Inflation: The Implications of
Analysts Earnings Forecasts, The Review of Economics and Statistics 84, pp. 63248.
Sharpe, W., 1964, Capital Asset Prices: A Theory of Market Equilibrium under Conditions
of Risk, The Journal of Finance 19, pp. 42542.
Solnik, B., 1983, The Relation between Stock Prices and Inflationary Expectations: The
International Evidence, The Journal of Finance 38, pp. 3548.
Wei, C., 2009, Does the Stock Market React to Unexpected Inflation Differently across the
Business Cycle?, Applied Financial Economics 19, pp. 194759.
Wei, C., 2010, Inflation and Stock Prices: No Illusion, Journal of Money, Credit, and Banking
42, pp. 32546.
323
i
i
i
perrucci 2012/7/11 17:12 page 324 #346
i
i
i
perrucci 2012/7/11 17:12 page 325 #347
14
i
i
i
perrucci 2012/7/11 17:12 page 326 #348
INFLATION-SENSITIVE ASSETS
investor, this implies that static hedges may not grant adequate protection. At short horizons, commodities, and to some extent cash,
provide some protection against the eroding effect of unanticipated
inflation. Over the long run, their efficiency fades away, and nominal bonds start to offer better real return features as higher yields
provide a greater income cushion. Like many others (see, for example, Bodie 1976; Jaffe and Mandelker 1976; Fama and Schwert 1977;
Solnik 1983), we found that nominal equity returns are negatively
correlated to inflation.
This chapter extends our earlier work on long-run dynamics and
focuses on two issues. The first is to identify the most effective inflation hedges, going beyond broad asset classes, and looking into a
range of US domestic and international fixed-income instruments
and equity sectors. The second is to study asset class hedge performance after an inflation shock, over different time horizons, in order
to gain insight into possible active asset allocation/sector rotation
strategies.
In the context of a diversified investment portfolio, we use a multivariate vector error-correction (VEC) model, and calculate impulse
responses, so as to assess how inflation shocks affect asset and portfolio returns over time. We find that long-term investors could benefit from sector rotation amongst different categories of bonds and
equities. As expected, these properties experience large variations
over time.
WHAT CAN WE LEARN FROM PREVIOUS RESEARCH?
The theory surrounding this subject suggests that some traditional
asset classes should provide relatively effective inflation protection.
American economist Irving Fisher suggested that the nominal return
on short-term debt should comprise a real interest rate and compensation for expected inflation (Fisher 1930). In other words, Treasury bills (T-bills), commonly referred to as cash, or the risk-free
rate, would provide a perfect hedge when inflation is anticipated.
That is, short-term real interest rates would remain unchanged. This
hypothesis was subsequently challenged by Mundell (1963) and
Tobin (1965), who both argued that nominal interest rates should
change by less than one-to-one with changes in expected inflation,1
reducing the effectiveness of cash as a hedge. Atti and Roache (2009)
also found that cash is an imperfect hedge against unanticipated
326
i
i
i
perrucci 2012/7/11 17:12 page 327 #349
i
i
i
perrucci 2012/7/11 17:12 page 328 #350
INFLATION-SENSITIVE ASSETS
three types of nominal bonds, ie, government bonds, mortgagebacked securities and corporate bonds. Furthermore, we analyse
10 global equity sectors, as well as international diversification
through foreign equities and bonds.
MODEL SPECIFICATION
Many investors set their strategic asset allocation and their investment decisions based on expected risk and return over a relatively long time period, often five years or more. To assess asset
class inflation-hedging properties over such horizons, we use a cointegrated vector autoregressive (VAR) approach that allows us to
both identify whether asset class returns and inflation share common trends over the long run, and assess their dynamics over the
short run.2 Specifically, our objective is to estimate how different
asset classes react to an unexpected rise in inflation. An inflation
surprise is defined here as an increase in the US Consumer Price
Index (CPI) that is not anticipated by the VAR model. Clearly, market participants, who have access to a larger information set, may
anticipate what might be a surprise for any specific stochastic model
employed. However, one of the strengths of the VAR approach is that
it encapsulates a lot of information already. It can be written as
Zt = +
P
p=1
p Z tp + t
(14.1)
i
i
i
perrucci 2012/7/11 17:12 page 329 #351
J
j =1
j j Z t1 +
P
1
p Z tp + t
(14.2)
p=1
i
i
i
perrucci 2012/7/11 17:12 page 330 #352
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 331 #353
In other words, inflation shocks affect all other variables, cash return
shocks affect all variables except inflation, bond return shocks affect
all other variables except inflation and cash, and so on.
Although the ordering of variables is somewhat arbitrary, our
empirical results turn out to be insensitive to this ordering, largely
reflecting the low correlation of the reduced-form residuals from the
estimated model.4
DATA
In our study, we use monthly data from a variety of domestic and
international total return series. Total return series, in addition to
price returns, assume interest and dividend earned are reinvested
in subsequent periods. Wherever possible, we selected indexes (as
detailed below) that are widely used by investors as performance
benchmarks.5 The data set spans a time window from January 1970
to April 2011 (with a few exceptions highlighted in the paragraph
below).
For inflation, we use the CPI (all urban consumers) published
monthly by the US Bureau of Labor Statistics. For cash, we use the 90day US Treasury bills total return index provided by Global Financial
Data.6 For nominal bonds, we use the Barclays US Aggregate Index.
We also analyse a few sub-indexes within investment grade bonds,
namely the Barclays US Treasury Index, the Barclays US Mortgage
Backed Securities (MBS) Index, and US corporate bonds (measured
by the Merrill Lynch Corporate Master Index). The data set from
these four nominal bond indexes starts from December 1975.
While global equity indexes are widely available for the entire
period under review, global aggregate bond indexes are available
over a much shorter time span (since about 1990 for the Barclays Global Aggregate index). For the purpose of this chapter, and
because of the breadth of reliable data in the US bond market, we
opted to measure the sensitivity to inflation shocks of US-only investment grade bonds. Possible discrepancies in the degree of responses
to those shocks are, in fact, relatively limited: the correlation between
the US and global bond indexes since 1990 is high at about 0.9 and
the impulse response results from the model are qualitatively similar and within 25bp of the response on US bond aggregates for
the same sample periods. This reflects the large share of US bonds
331
i
i
i
perrucci 2012/7/11 17:12 page 332 #354
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 333 #355
US inflation
US T-bill
Bond indexes:
Barclays US
Aggregate
Barclays US MBS
Barclays US Treasury
Merrill Lynch/BoA
Corporate Master
JP Morgan EMBI+
JP Morgan ELMI+
MSCI equity indexes:
World
World value
World growth
Energy
Materials
Industrials
Consumer
discretionary
Consumer staples
Health care
Financials
IT
Telecoms
Utilities
GSCI Commodity
Total Return Index
MSCI Emerging
Markets
SD
0.4
0.5
1.8 1.8
1.3
0.0
1.2
0.9
0.0
0.4
0.06
0.10
0.15
0.28
0.7
10.8 6.3
5.6
0.5
0.58
0.00
0.7
0.7
0.7
14.5 7.9
9.2 5.1
11.3 7.7
6.6
5.5
6.9
0.9
0.3
0.0
0.81
0.43
0.76
0.00
0.00
0.00
0.8
0.7
0.90
0.95
0.00
0.00
0.8
1.0
0.8
1.0
0.8
0.8
0.7
13.7 21.0
14.4 20.5
14.0 21.5
16.6 23.4
16.9 31.0
15.6 24.7
17.2 19.5
15.1
14.9
15.9
18.3
19.5
17.2
17.1
0.8
0.8
0.7
0.4
0.8
1.0
0.6
0.72
0.45
0.61
0.93
0.89
0.80
0.84
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.8
0.9
0.8
0.7
0.8
0.9
0.8
16.3 19.5
19.0 18.7
20.9 29.8
21.4 31.2
26.8 17.3
22.5 14.3
22.9 33.1
14.9
14.6
19.7
22.7
17.5
14.5
19.9
0.7
0.4
0.6
0.6
0.1
0.0
0.4
0.93
0.86
0.68
0.83
0.65
0.80
0.26
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.8
0.5
0.21
0.00
then the ELMI+ and finally the MSCI Emerging Market Equity Index.
The impulse responses of the core variables from these three models
333
i
i
i
perrucci 2012/7/11 17:12 page 334 #356
INFLATION-SENSITIVE ASSETS
Inflation
Cash
Bonds
Equities
Commodities
6 months
12 months
2 years
Long run
1.65
(0.33)
0.01
(0.20)
0.97
(0.04)
1.65
(0.17)
5.59
(0.51)
1.70
(0.49)
0.15
(0.43)
1.13
(0.28)
2.38
(0.60)
4.24
(1.03)
1.94
(1.81)
0.58
(0.56)
0.96
(0.57)
2.64
(1.85)
3.56
(2.06)
2.17
(1.01)
1.45
(1.11)
0.03
(1.07)
2.20
(3.35)
3.65
(3.96)
Bootstrapped standard errors are given in parentheses. The data represents cumulative total log change for the CPI, and cumulative total rate of
return for the four core asset classes.
are the average from each of the separately estimated models, which
tended to be qualitatively similar.
Table 14.2 shows the results, specifically the cumulative total
change for the CPI, and the cumulative total rate of return for the four
core asset classes. For illustration, four time horizons (six months,
one year, two years and ten years (labelled as long run)) are shown
in the table.
US inflation: shocks persist
As expected, US inflation exhibits strong autoregressive properties.
In fact, an initial month-on-month shock of 1% to the CPI causes
an increase in subsequent months, with an effect felt long in the
future. After one year, the cumulative increase in CPI is 70% higher
than the initial annualised shock. Over a 10-year horizon, consumer
prices have risen a cumulative 2.17%, ie, more than twice the initial
shock.
US cash: a far-from-perfect inflation hedge
As a proxy for US dollar denominated cash, we use three-month US
Treasury bills. Given that the short-end of the yield curve is most
directly affected by monetary policy actions, cash returns should
increase in response to an inflation surprise, if there is an expectation
of tighter monetary policy. However, as shown in Table 14.2 and
Figure 14.1, the response is gradual, and the hedge is far from perfect.
334
i
i
i
perrucci 2012/7/11 17:12 page 335 #357
One year from the shock, consumer prices have increased by 1.70%
(see the 12-month column in Table 14.2), while cash returns are
only 0.15%. In the long run, the cumulative return from cash reaches
1.45% (see the long run column in Table 14.2), but still does not
compensate for the loss in purchasing power up to that point (2.17%).
US aggregate bonds: negative real returns
The inverse relationship between US nominal bonds (measured by
the Barclay US Aggregate Bond Index) and US inflation is clear, as the
purchasing power of fixed nominal coupons and notional payments
naturally diminishes with rising inflation. Indeed, following an inflation shock, bonds underperform in comparison to cash. Among the
securities in the aggregate index, long-duration bonds are obviously
the worst performing (due to their long duration). Six months after
the initial consumer prices shock, cumulative inflation has reached
1.65% (see the 6-month column in Table 14.2), while the US aggregate bond index has lost 0.97% (ie, a 2.62% real return). There is
limited recovery over time: at year 10, the US Aggregate Bond Index
reaches a flat cumulative nominal return, but its real return remains
deep in negative territory (2.14%).
In the next section, we analyse the impulse response function
of other inflation-hedging alternatives; specifically, instead of an
aggregate bond index, we consider sector indexes (US treasuries, US
mortgage-backed securities and US corporate bonds) and compare
their performance following an initial consumer prices shock.
Developed markets equities: the worst performing core asset
class hedge
Developed market equity returns (both nominal and real) are deep
in negative territory for all time horizons considered. In the long
run, nominal returns are 2.20% (long run column in Table 14.2),
while inflation-adjusted real returns are 4.37%. These findings are
in line with our earlier work (Atti and Roache 2009), and add further
evidence to other empirical observations, which have questioned the
theoretical underpinnings of equities as an inflation hedge (because
of their real asset characteristics).
In the next section, we shall extend the impulse response analysis
to non-core asset classes, and consider the inflation-hedging performance of the 10 GSCI equity sectors (Table 14.1), value versus growth
equity strategies, and emerging market equities.
335
i
i
i
perrucci 2012/7/11 17:12 page 336 #358
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 337 #359
2.0
1.6
1.2
0.8
0.4
0
2.5
2.0
1.5
1.0
0.5
0
0 1 2 3 4 5 6 7 8 9 10
(c)
1.0
(b)
0 1 2 3 4 5 6 7 8 9 10
(d)
Treasuries
0
1.0
2.0
1.0
MBS
Corporates
2.0
3.0
0 1 2 3 4 5 6 7 8 9 10
(e)
4
2
0
2
2
6
Energy
Utilities
0 1 2 3 4 5 6 7 8 9 10
Years
6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10
(f)
0 1 2 3 4 5 6 7 8 9 10
Years
(a) Inflation; (b) US three-month T-bills; (c) US bonds; (d) MSCI World Equity Index;
(e) MSCI world equity sector indexes; (f) S&P GSCI.
i
i
i
perrucci 2012/7/11 17:12 page 338 #360
INFLATION-SENSITIVE ASSETS
US Treasuries
US MBS
US Corporates
6 months
12 months
2 years
Long run
0.59
(0.05)
1.13
(0.15)
1.80
(0.21)
0.41
(0.11)
1.08
(0.36)
1.90
(0.41)
0.23
(0.22)
1.08
(0.72)
1.96
(0.82)
0.05
(0.42)
0.72
(1.34)
2.26
(1.44)
with the Merrill Lynch US Corporate Master Index. The results are
shown in Table 14.3.
As can be seen in Table 14.3, mortgage-backed securities and corporate bonds underperform Treasuries following an inflation shock.
The impulse response for Treasury performance is taken from the
mortgage-backed securities model (the results for Treasuries from
both models, including the standard errors, were almost identical).
In the case of mortgages, their negative convexity makes mortgage-backed securities particularly sensitive to rising real yields. As
homeowners are less likely to prepay, the value of the embedded
optionality in mortgage-backed securities decreases and lengthens
the residual duration of the bond. In the case of corporate bonds, performance is affected in a similar fashion to the case of traded equities:
higher real yields weigh on firms borrowing costs and future overall
profitability.
Over the long run, once the effect of the shock is fully priced in by
investors, Treasuries recover with the help of higher coupons (and
current yields). However, as seen in Figure 14.1, their real cumulative
return remains negative, owing to large early losses that are never
fully recouped.
Developed market equities: sector indexes
After analysing the aggregate global equity index in the previous
section, here we look at the inflation-hedging performance of the
10 traditional Global Industry Classification Standard (GICS) sectors
distinguished by MSCI (Table 14.1).
We estimate 10 different six-variable VAR models with US CPI,
US cash, US bonds, global equities and commodities as the fixed
338
i
i
i
perrucci 2012/7/11 17:12 page 339 #361
Energy
Materials
Industrials
Consumer
discretionary
Consumer
staples
Health care
Financials
IT
Telecoms
Utilities
6 months
12 months
2 years
Long run
3.14
(0.51)
2.26
(0.67)
2.78
(0.57)
3.40
(0.51)
3.70
(0.36)
2.42
(0.46)
0.05
(0.67)
2.52
(0.67)
3.09
(0.62)
5.25
(0.51)
1.03
(1.08)
2.73
(1.34)
2.62
(1.18)
4.22
(1.03)
3.86
(0.77)
2.73
(0.93)
1.44
(1.39)
3.70
(1.39)
3.09
(1.23)
4.94
(1.03)
0.10
(2.16)
3.24
(2.73)
2.47
(2.37)
4.48
(2.11)
4.01
(1.60)
2.57
(1.85)
1.90
(2.83)
3.91
(2.83)
3.04
(2.47)
4.58
(2.06)
0.93
(3.76)
4.68
(4.89)
3.19
(4.27)
4.89
(3.81)
4.22
(2.88)
1.13
(3.60)
1.96
(5.15)
4.01
(5.15)
3.09
(4.42)
4.12
(3.96)
variables, and each of the 10 GICS sectors rotating in and out of the
model in turn.
The first key finding is that the poor performance of developed
markets equities following an inflation shock is broadly spread
across sectors, and occurs rapidly (Table 14.4). Out of the 10 traditional GICS sectors, 9 register a negative impulse response at every
time period following the shock. This adverse effect on returns follows very quickly after the initial shock. For example, total returns
for the Industrials sector are 2.78% six months after the initial
inflation shock, and remain negative thereafter. A similar profile is
evident for most other global equities sectors.
The second key result is that there is noticeable difference in
performance across sectors. Unsurprisingly, following an inflation
shock, the energy sector has positive nominal returns in the short
run, although this effect appears to dissipate over time.
339
i
i
i
perrucci 2012/7/11 17:12 page 340 #362
INFLATION-SENSITIVE ASSETS
This partly reflects the important role that oil prices play in inflation shocks, and underscores the importance of understanding the
underlying cause of higher inflation. However, the materials sector,
which includes mining companies, fails to offer similar short-run
inflation protection. This may change, as commodity prices have
become more positively correlated with oil prices. Perhaps surprisingly, financials have tended to perform relatively well in the aftermath of inflation shocks (Table 14.4), although high standard errors
indicate that this is not a particularly robust result. In the short run,
the worst performer is the utilities sector, which underperforms significantly in the first year following the shock. Again, this may reflect
the nature of the shock, as utilities use inputs whose prices are closely
linked to those of crude oil, including coal and natural gas. It is also
due to the frequently regulated nature of pricing which often prevents higher input costs from being passed to consumers, thus negatively impacting profitability. Furthermore, leverage might play a
part, as utilities often have high levels of debt, reflecting the relatively more stable nature of their business. Firms in sectors with less
stable revenues and profits cannot achieve comparable leverage in
the market as they are perceived as more risky. Clearly, an increase in
interest rates and debt servicing costs has a negative impact on their
overall profitability. In the long run, all equities sectors post negative nominal, and inflation-adjusted, returns, with materials and
consumer discretionary being the worst performing indexes.
Developed market equities: growth versus value styles
We next assessed the inflation-hedging features of global equities
across different investment styles (Table 14.5). Again, our two sixvariable models consisted of five core variables (US inflation, US
cash, US bonds, developed market equities, commodities), with one
model including as our rotating variable a global equity growth style
index, and the other a global equity value style index.
Albeit with a large standard error, results show that, following
an inflation shock, growth stocks perform worse than value stocks
(Table 14.5). One reason for this could be that growth stocks are long
duration, with a larger portion of the dividend stream occurring
further in the future. Therefore, they are more sensitive to a rise
in inflation, and consequent high interest rates, than value stocks,
which typically offer a higher current dividend yield.
340
i
i
i
perrucci 2012/7/11 17:12 page 341 #363
12 months
2 years
Long run
2.21
(0.51)
2.93
(1.08)
3.24
(2.16)
3.45
(3.81)
1.80
(0.46)
2.21
(0.98)
2.32
(1.96)
2.47
(3.45)
12 months
2 years
Long run
EM bonds (EMBI+)
3.50
(0.36)
3.76
(0.77)
1.65
(1.60)
2.11
(2.78)
EM bonds (ELMI+)
1.97
(0.27)
1.65
(0.77)
1.40
(0.56)
4.01
(1.60)
2.90
(1.14)
2.98
(3.29)
3.01
(1.99)
1.23
(5.92)
Emerging market
equities
i
i
i
perrucci 2012/7/11 17:12 page 342 #364
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 343 #365
Benchmark
130
Portfolio performance
125
Second
rebalancing
120
Back to
neutral
115
110
105
First
rebalancing
100
95
90
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of months
The vertical axis shows the portfolio index value, which starts at 100.
i
i
i
perrucci 2012/7/11 17:12 page 344 #366
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 345 #367
CONCLUSIONS
This chapter explored the long-run dynamics of different assets following a US inflation shock. We were able to identify the most effective inflation hedges, going beyond broad asset classes, and looking into a range of US domestic and international fixed-income and
equity sectors.
Our first set of results compared the impulse response of our five
core variables (CPI and four core asset classes, ie, US cash, US aggregate bonds, developed market equities, and commodities) following
a 1% shock to the monthly CPI. As expected, US inflation exhibited strong autoregressive properties; commodities were the best
performing class, while equities underperformed.
Following the analysis of asset classes at the aggregate level, we
analysed the impulse response at a more disaggregated level, detailing our results along the way. Hedge performance was analysed at
different time horizons, in order to gain insight into possible active
asset allocation/sector rotation strategies.
What are the practical implications of these findings? First, traditional broad asset classes provide an imperfect inflation hedge,
at best, and in some important cases, including equities, do very
poorly. But, for long-term long-only investors, there is still hope.
In particular, for those with confidence in their views about the
path of future inflation, there is scope to enhance inflation protection through tactical asset allocation and sector rotation within each
asset class. Even within equities, there is broad divergence across
sectors in their inflation-hedging properties. Similar divergence is
evident within the bond universe. In summary, investors need to
look beyond broad asset classes and be willing to take sectoral bets
to ensure optimal inflation protection.
345
i
i
i
perrucci 2012/7/11 17:12 page 346 #368
INFLATION-SENSITIVE ASSETS
The Taylor Rule, which is a descriptive model of how the US Federal Reserve have been
setting the funds rate since the early 1990s, empirically contradicts this, as a 1% change in
inflation translates into a higher percentage change in the short-term rate.
Granger and Newbold (1974) and Johansen (1991) are the seminal papers describing these
statistical techniques.
The hypothesis that the first differences follow a unit root process can be rejected at the 5%
level for all variables, except for the Treasury bill total return index, for which rejection is
possible only at the 10% level.
See Atti and Roache (2009) for further details on model specifications.
Bailey (1992) lists six qualities of a valid benchmark: unambiguous, investable, measurable,
appropriate, reflective of current investment opinions and specified in advance.
See http://www.globalfinancialdata.com.
See http://www.msci.com.
REFERENCES
Akaike, H., 1974, A New Look at the Statistical Model Identification, IEEE Transactions
on Automatic Control 19(6), pp. 71623.
Ang, A., M. Brire and O. Signori, 2011, Inflation and Individual Equities, URL: http://
ssrn.com/abstract=1805525.
Atti, A., and S. Roache, 2009, Inflation Hedging for Long-Term Investors, IMF Working
Paper WP/09/90.
Bailey, J. V., 1992, Are Manager Universes Acceptable Performance Benchmarks?,
Journal of Portfolio Management 18(3), pp. 913.
Balduzzi, E., and C. Green, 2001, Economic News and Bond Prices: Evidence from the
US Treasury Market, Journal of Financial and Quantitative Analysis 36, pp. 52343.
Bekaert, G., and X. Wang, 2010, Inflation Risk and the Inflation Risk Premium, Economic
Policy, October, pp. 755806.
Bodie, Z., 1976, Common Stocks as a Hedge against Inflation, The Journal of Finance
31(2), pp. 45970.
Brire, M., and O. Signori, 2010, Inflation-Hedging Portfolios in Different Regimes,
Working Paper 5, Amundi.
Buckland, R., 2008, The Inflation Threat, Citi Global Equity Strategist, July 5.
Dimson, E., P. Marsh and M. Staunton, 2011, Credit Suisse Global Investment Returns
Yearbook, CSFB Research Institute.
Evans, M. D. D., 1998, Real Rates, Expected Inflation, and Inflation Risk Premia, The
Journal of Finance 53(1), pp. 187218.
Fama, E. F., and G. W. Schwert, 1977, Asset Returns and Inflation, Journal of Financial
Economics 5(2), pp. 11546.
Fisher, I., 1930, The Theory of Interest (New York: Macmillan).
Gorton, G, and G. K. Rouwenhorst, 2006, Facts and Fantasies about Commodity
Futures, Financial Analysts Journal 62(2), pp. 4768.
346
i
i
i
perrucci 2012/7/11 17:12 page 347 #369
347
i
i
i
perrucci 2012/7/11 17:12 page 348 #370
i
i
i
perrucci 2012/7/11 17:12 page 349 #371
Part III
i
i
i
perrucci 2012/7/11 17:12 page 350 #372
i
i
i
perrucci 2012/7/11 17:12 page 351 #373
15
Nic Johnson
PIMCO
There are many different schools of thought as to what causes inflation, and over the years many explanations have been put forward to explain historically observed inflation dynamics. There are
monetarists, who believe that changes in-the-money supply are the
most important inflationary dynamic, while Keynesian economists
argue that underlying pressures in the economy play a large role in
determining inflation levels.
In this chapter, we shall not delve into the academic and philosophical issues surrounding inflation but instead focus on the
forecasting methods used by practitioners, including top-down,
bottom-up and time-series-based models. We shall look at the appropriateness of different frameworks, depending on the type of data
inputs available, as well as the frequency and length of forecast
desired. Finally, we shall consider some of the differences involved
with modelling inflation in the US, and other developed global
markets, relative to emerging markets.
FUNDAMENTAL TOP-DOWN MODELS FOR FORECASTING
INFLATION
The goal of a top-down macro inflation model is to identify economic variables that are leading indicators of changes in the price
of various goods and services. From these leading indicators, an
investor will be able to better predict both the direction and the
magnitude of changes in the rate of inflation. Top-down models are
mostly used for analysing core inflation, since the prices of food and
351
i
i
i
perrucci 2012/7/11 17:12 page 352 #374
INFLATION-SENSITIVE ASSETS
19972010
19851993
19751984
14
12
10
8
6
4
2
0
0
4
6
8
Unemployment rate (%; lagged 1 yr)
10
12
i
i
i
perrucci 2012/7/11 17:12 page 353 #375
i
i
i
perrucci 2012/7/11 17:12 page 354 #376
INFLATION-SENSITIVE ASSETS
Figure 15.2 Chicago Fed National Activity Index and core inflation
(19972010)
3.0
2.5
2.0
1.5
1.0
0.5
0
3.0
2.5
2.0
1.5
1.0
0.5
0.5
1.0
i
i
i
perrucci 2012/7/11 17:12 page 355 #377
14
12
10
8
6
4
2
0
4
6
8
10
12
14
Year-over-year change M2 (%; lagged 3 years)
16
i
i
i
perrucci 2012/7/11 17:12 page 356 #378
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 357 #379
i
i
i
perrucci 2012/7/11 17:12 page 358 #380
INFLATION-SENSITIVE ASSETS
Categories
Weight
(%)
Examples of
leading indicators
Shelter
31.96
14.79
Medical
6.63
Recreation
6.29
5.57
Motor fuel
5.08
Home furnishing
4.41
Household energy
4.00
Apparel
3.60
Communication
3.31
Education
3.11
Personal care
2.59
Public transport
1.23
7.45
i
i
i
perrucci 2012/7/11 17:12 page 359 #381
Sector
Volatility
contribution (%)
Petroleum
Shelter
Household energy
Food
New and used cars
Recreation
Communication
Home furnishing
Public transport
Apparel
Medical
Education
0.98
0.35
0.27
0.17
0.13
0.06
0.06
0.05
0.05
0.04
0.04
0.02
Core
Headline
0.43
1.16
i
i
i
perrucci 2012/7/11 17:12 page 360 #382
INFLATION-SENSITIVE ASSETS
this price information is readily available and transparent and represents an actual market price that can be used for hedging. The
spot price of petroleum reflects all of the currently known supply
demand factors, both existing and those expected in the future. The
future prices are linked to the spot price through an arbitrage condition, ie, through the addition of financing and storage costs and
a convenience yield. However, the extent to which the futures price
of petroleum is an accurate or reliable forecast of prices in the future
is questionable, considering that realised and implied volatility are
generally quite high. For example, the typical annual implied volatility of options on petroleum futures is around 3040%, which points
to the intrinsic difficulty in obtaining an accurate forecast at comparable time horizons. Nevertheless, without a fundamental model of
supply and demand, the futures curve remains the common starting
point for petroleum forecasting in the medium to long term.
Forecasting in the very short term, specifically focusing on the
next petroleum inflation number, is actually easier, the reason being
twofold:
1. the inflation index is published at a monthly frequency,
while information on price behaviour is observed at higher
frequency (for example, daily);
2. there is a one-month lag, as the CPI published in month n is a
measure of inflation in month n 1.2
High-frequency price information can come from various sources.
These include daily petroleum futures prices and daily retail market
surveys such as the one conducted by the Automobile Association
of America (AAA). Each of these daily data sources can then be averaged over month n 1 to accurately forecast the upcoming rate of
petroleum inflation. Note, however, that using petroleum futures
prices involves a fair amount of basis risk since the contract represents wholesale petroleum prices with delivery in New York Harbor,
while the petroleum inflation in the CPI is an average of retail prices
across the US. As a result of this basis risk in the petroleum future,
the AAA petroleum price survey is the most accurate way to forecast petroleum inflation in the very short term, since it represents
retail prices from gas stations across the country. Not surprisingly,
the AAA petroleum survey displays an impressive 99% correlation
with the monthly petroleum CPI inflation rate.
360
i
i
i
perrucci 2012/7/11 17:12 page 361 #383
Shelter inflation
After petroleum, the next most important sub-index in the CPI, when
it comes to volatility contribution, is shelter. Shelter has two main
components: the rent of primary residence (about 6% of the CPI)
and the owners equivalent rent of primary residence (OER, about
23% of the CPI). There is a substantial amount of misinformation
about OER, but the main point is that, for both the rent and the
OER components, the US Bureau of Labor Statistics (BLS) uses actual
rather than imputed3 rent data (collected in the CPI Housing Survey)
to calculate inflation rates. Hence, for both components, the main
variable to model is the rate of inflation of the average consumers
rent.
As mentioned before, some of the economic factors that could
be important to forecasting changes in rental rates are those that
describe the supplydemand balance of rentals, and those that compare the relative cost of renting versus buying a home. In addition
to these macro factors, there is also a microeconomic effect that is
important for higher frequency models, namely utility price adjustments. Since most rental rates include some utility payments, if, for
example, the price of water goes up sharply in one month and water
is included in the rent, then the rent is effectively decreased. The
same dynamic is true for other utility prices, such as electricity or
natural gas. The BLS attempts to compensate for these effects. Therefore, when utility prices increase, this places downward pressure on
shelter inflation, and vice versa when utility prices fall. A bottom-up
model can capture these effects by feeding the output of a household
energy and utilities model, discussed in the following section, into
a model for shelter inflation.
Household energy inflation
Household energy consists primarily of retail power prices. A logical
place to start is wholesale power prices, as these are available across
Europe, the US and several other countries. By examining this data, it
is possible to establish the relationship between wholesale and retail
prices, including the relative sensitivity (beta) and the leadlag.
If the lead time from wholesale to retail power prices is shorter
than the time horizon sought for the forecast (say, for example,
wholesale prices have a three-month lead to retail prices, but a
twelve-month forecast of retail prices is desired), then we could look
361
i
i
i
perrucci 2012/7/11 17:12 page 362 #384
INFLATION-SENSITIVE ASSETS
Food inflation
70
60
50
40
30
20
10
0
10
20
30
40
Year-over-year GSCI
Agricultural and Livestock Index (%)
7
6
5
4
3
2
1
0
1
Dec 1991
Dec 1992
Dec 1993
Dec 1994
Dec 1995
Dec 1996
Dec 1997
Dec 1998
Dec 1999
Dec 2000
Dec 2001
Dec 2002
Dec 2003
Dec 2004
Dec 2005
Dec 2006
Dec 2007
Dec 2008
Dec 2009
Dec 2010
Dec 2011
Figure 15.4 Food inflation and agricultural and livestock price changes
i
i
i
perrucci 2012/7/11 17:12 page 363 #385
i
i
i
perrucci 2012/7/11 17:12 page 364 #386
INFLATION-SENSITIVE ASSETS
(a)
y = 0.9421x
0.6
R2 = 0.5448
0.5
0.4
0.3
0.2
0.1
0
0.1
0.2
0.2
0.1
0.1
0.2
0.3
0.4
0.5
0.6
y = 0.5456x
0.3
R2 = 0.3005
0.2
0.1
0
0.1
0.2
0.3
0.3
0.2
0.1
0.1
0.2
0.3
0.4
OER month-over-month one-month lag minus OER MoM two-month lag (%)
(a) Trending nature; (b) mean-reverting nature.
Source: data from PIMCO, Bloomberg and BLS from 2000 to 2011.
i
i
i
perrucci 2012/7/11 17:12 page 365 #387
Commodity
pass-through
to food CPI (%)
Commodity
pass-through to
headline CPI (%)
US
EU countries
UK
Japan
Canada
Australia
13.7
14.0
9.3
25.9
17.0
15.4
6.2
5.9
15.6
6.0
6.4
2.6
0.8
0.8
1.5
1.6
1.1
0.4
Average
15.9
7.1
1.0
China
Brazil
Mexico
India
Russia
Turkey
33.0
30.2
22.7
47.1
38.0
27.6
25.0
14.7
9.0
16.2
16.2
10.0
8.2
4.4
2.0
7.6
6.1
2.8
Average
33.1
15.2
5.2
Country
of both bottom-up- and top-down-type models. Furthermore, correctly modelling seasonal factors is important because the returns
that an investor earns from holding US Treasury Inflation-Protected
Securities are linked to the non-seasonally adjusted level of inflation.
To conclude, technical analysis is useful in studying the time series
characteristics of inflation indexes, including seasonality patterns,
and thus it can add to the accuracy of both top-down and bottom-up
inflation models, particularly at higher frequencies.
FORECASTING INFLATION IN DEVELOPED VERSUS
EMERGING MARKETS
Although similar methods and principles can be used to forecast
inflation in both developed and emerging economies, there are some
noteworthy differences.
Emerging markets (EMs) tend to have a much higher food weighting in their consumption baskets, while developed markets (DMs)
tend to have higher weightings of services. Table 15.3 shows the
weightings of food in the CPI basket for a handful of emerging and
365
i
i
i
perrucci 2012/7/11 17:12 page 366 #388
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 367 #389
economic data are plentiful, a bottom-up inflation model can complement a top-down approach, increase short-term accuracy and
give early warning of structural shifts in the underlying inflation
dynamics. In practice, building such a model can prove more challenging for emerging economies, where available data might be
relatively sparse.
1
Commodity spot and future prices typically exhibit strong seasonality (because either demand
or supply is seasonal). The forward price at time t for delivery at T, F(t, T ), inherits the
seasonality of the underlying spot price P (t) at time t. Consequently, to extract a forecast of
future price at time T, we need to adjust for seasonality effects.
Of course, we have to bear in mind that market prices also adjust to the same high-frequency
information, so this inflation forecast is likely to be of limited use in trading ahead of the
curve.
Part of the confusion arises because the relative weight of OER in the CPI (about 23%) is
calculated from imputed expenditures, derived from asking a sample of homeowners the
question: If someone were to rent your home today, how much do you think it would rent
for monthly, unfurnished and without utilities?.
REFERENCES
Fisher, I., 1926, A Statistical Relation between Unemployment and Price Charges, International Labour Review 13(6), pp. 78592. (Reprinted as I Discovered the Phillips Curve:
A Statistical Relation between Unemployment and Price Changes , Journal of Political
Economy 81(2), pp. 496502 (1973).)
Friedman, M., 1963, Inflation: Causes and Consequences (New York: Asia Publishing House).
Phillips, A. W., 1958, The Relationship between Unemployment and the Rate of Change
of Money Wages in the United Kingdom 18611957, Economica 25(100), pp. 28399.
Stock, J., and M. Watson, 1999, Forecasting Inflation, Journal of Monetary Economics 44(2),
pp. 293335.
367
i
i
i
perrucci 2012/7/11 17:12 page 368 #390
i
i
i
perrucci 2012/7/11 17:12 page 369 #391
16
i
i
i
perrucci 2012/7/11 17:12 page 370 #392
INFLATION-SENSITIVE ASSETS
Inflation (%)
16
14
12
10
8
6
4
2
0
2
4
1950
1960
1970
1980
1990
2000
2010
Inflation (%)
Mean survey of
professional forecasters
5Y moving average
US CPI-U inflation
3
2
1
0
1
2
12M US CPI-U
inflation
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
i
i
i
perrucci 2012/7/11 17:12 page 371 #393
Figure 16.3 Cumulative inflation: (a) actual versus expected and (b) five
year cumulative difference between expected and actual inflation
(a)
800
700
Expected inflation
(cumulative since 1957)
600
Actual inflation
(cumulative since 1957)
500
% 400
300
200
100
0
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
(b)
35
30
25
20
15
% 10
5
0
5
10
15
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
i
i
i
perrucci 2012/7/11 17:12 page 372 #394
INFLATION-SENSITIVE ASSETS
2020
2010
2000
1990
1980
1970
1960
1950
1940
1930
1920
2,800
2,400
2,000
1,600
1,200
800
400
0
400
1910
US dollars (billions)
i
i
i
perrucci 2012/7/11 17:12 page 373 #395
their economies. The extended period of low inflation in the postVolcker era might be contributing to a false sense of security, while
increasing the likelihood of underestimating inflation in the future.
A reasonable level of inflation in line with expectations is generally not a big threat for businesses and consumers, as nominal
prices including output prices and wages should adjust accordingly.
However, the insidious risk is an unexpected change in the rate
of inflation, especially in the environment at the time of writing,
where there are reasons to believe traditional measures of inflation
expectations might be misleading. Therefore, inflation risk should
be carefully analysed, as it can negatively affect both consumers and
businesses. This issue is particularly acute for property and casualty
(P&C) insurers: a topic discussed in the next section.
INFLATION RISK FOR PROPERTY AND CASUALTY
COMPANIES
Property and casualty (P&C) insurers are exposed to a wide variety
of risks that must be managed within a companys risk tolerance.
For example, the risk of catastrophic weather events can be managed
through property location diversification or reinsurance, while other
risks are more difficult to protect against.
Among such challenging exposures is the risk of unexpected
(higher) inflation, which tends to be an insidious peril that grows
over time and affects all key financial metrics of an insurer. On the
liability side of the balance sheet, claim payments accelerate beyond
expected levels and additional reserves are required in anticipation
of higher claims, often leading to poor underwriting results as liabilities exceed incoming cashflow streams. Company expenses also
escalate beyond expectations.
The magnitude of the impact on a companys performance
depends on a host of factors. Obviously, first in line is the impact of
inflation on liability payments. But other issues are also important,
such as the ability to pass higher costs (due to higher inflation) to the
customers, the magnitude of the upside run in prices and how long
the inflationary period lasts. Finally, the adverse impact on reserve
adequacy is more manifest on liabilities further in the future, as inflation surprises have a longer time to cumulate, thus increasing cashflows payable. Clearly, it is not just liabilities that are exposed to an
inflation surprise, as higher interest rates will also affect negatively
373
i
i
i
perrucci 2012/7/11 17:12 page 374 #396
INFLATION-SENSITIVE ASSETS
Underwriting
analysis
Asset analysis
Financial
statement
projections
Financial
analysis and
reports
Source: Conning Risk & Capital Management Solutions.
i
i
i
perrucci 2012/7/11 17:12 page 375 #397
i
i
i
perrucci 2012/7/11 17:12 page 376 #398
INFLATION-SENSITIVE ASSETS
Allocation (%)
11.0
31.9
20.6
5.6
4.5
4.1
1.9
12.1
1.5
6.9
100.0
i
i
i
perrucci 2012/7/11 17:12 page 377 #399
Lines of business
Private passenger auto: liability
Private passenger auto: physical damage
Homeowners
Commercial auto: liability
Commercial auto: physical damage
Workers comp.
Commercial multiple peril: liability
Commercial multiple peril: property
Other liability
Products liability
Medical malpractice
Fire
Allied lines
Inland marine
Reinsurance
All other lines
16.0
0.8
3.7
4.5
0.1
20.8
4.8
1.8
21.3
2.9
5.2
0.8
0.9
0.6
7.7
8.2
Total
100.0
60
50
48
43
40
29
30
23
20
21
15 13
8
10
0
One-year horizon
Five-year horizon
i
i
i
perrucci 2012/7/11 17:12 page 378 #400
INFLATION-SENSITIVE ASSETS
700
3.0
2.5
650
2.0
600
1.5
550
1.0
500
450
0.5
2010
2011
2012
2013
2014
2015
i
i
i
perrucci 2012/7/11 17:12 page 379 #401
i
i
i
perrucci 2012/7/11 17:12 page 380 #402
INFLATION-SENSITIVE ASSETS
0.13
0.11
0.09
0.07
0.05
0.03
0.01
0.01
0.03
0.05
1950
M1
1960
M1
1970
M1
1980
M1
1990
M1
2000
M1
2010
M1
i
i
i
perrucci 2012/7/11 17:12 page 381 #403
There were, however, certain asset classes that generated superior performance relative to the overall market. Hard assets sectors with exposure to commodities used in industrial production,
as well as precious metals, generally outperformed during inflationary periods. Energy sectors and energy commodities such as
petroleum-based products were also strongly positively correlated
with inflation, and were often previously the underlying cause of
higher inflation. Exposure to these hard asset sectors and commodities can be gained through the futures market, financial vehicles
such as exchange-traded funds (ETFs) or direct equity investments
in companies that benefit from commodity price increases. Other
select sectors, such as real estate, retailing, materials and technology
also provided solid returns. Convertible bonds in inflation-sensitive
sectors are a capital-efficient way for insurers to gain equity-like
exposure while still maintaining a relatively high level of investment
income.
Other financial vehicles are directly linked to the inflation rate
itself. Treasury Inflation Protected Securities (TIPS), are indexed to
the US CPI and accrue additional principal at the rate of inflation,
although, in a low inflation environment, TIPS will produce lower
investment income than most other fixed-income investments. Inflation swaps also can provide inflation protection, as a fixed rate of
interest is paid in return for the rate of inflation realised over the
period of the swap. No cash outlay is required, but insurers will need
to meet certain regulatory standards (such as filing a Derivative Use
Plan, which is a legal requirement in many US states) before executing an agreement, and would be subject to the swap mark-to-market
volatility.
Floating-rate notes can also provide inflation protection, as the
underlying rate should adjust to a higher level along with inflation.
When additional credit or liquidity spreads are present, these might
also provide yield enhancement and extra income.
INFLATION-HEDGING ASSETS AND EFFICIENT FRONTIER
Based on the historical analysis of long-term dynamics of inflation,
economic growth, interest rates and returns of different asset classes,
we calibrated the economic and capital market simulation model to
provide realistic scenarios for economic indexes and asset returns.
381
i
i
i
perrucci 2012/7/11 17:12 page 382 #404
INFLATION-SENSITIVE ASSETS
11
10
US equities
Convertibles
Infl.
equities
High yield
Commodities
MBS
6
5
EAFE
FRN
Hedge funds
4
3
Cash
2
1
Gov.
Corp.
Muni.
TIPS
20
25
30
35
0
0
10
15
40
i
i
i
perrucci 2012/7/11 17:12 page 383 #405
F
Add inflation swap
Add other inflation assets
Add TIPS
Baseline
PC industry
B
A
AJ denote different efficient portfolios.Source: Conning Risk & Capital Management Solutions analytics.
Shareholder's equity
(US$ billions)
Baseline
Inflationary shock
Shock with inflation assets
2010
2011
2012
2013
2014
2015
Figure 16.10 shows several efficient frontiers for our hypothetical company, using the asset and liabilities profile of the average US P&C insurer. Each of the four lines in the chart is a plot
of the average and standard deviation of economic value of the
company at the five-year time horizon for a given set of instruments (weights varying depending on the riskreturn preferences)
allowed in the investment portfolio. The baseline is the efficient
frontier using the set of assets in Table 16.1. Note that the average weights in Table 16.1 do not correspond to an efficient portfolio (PC industry point in Figure 16.10). To this set (Table 16.1) of
investable assets, we add first US TIPS, then floating-rate notes,
commodities, inflation-hedging equities and convertibles; finally,
an inflation swap overlay is included. The isolated point in Figure 16.10 represents the average economic value of the insurer before
383
i
i
i
perrucci 2012/7/11 17:12 page 384 #406
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 385 #407
i
i
i
perrucci 2012/7/11 17:12 page 386 #408
INFLATION-SENSITIVE ASSETS
Investment income
Some inflation-sensitive investments, such as commodities and
equities, do not pay coupons, or their dividend income is uncertain. Others, such as TIPS, have a lower fixed coupon yield than
most other fixed-income securities (although inflation might offset this). Therefore, the inflation investment hedging strategy might
decrease book yield in nominal terms, although it should result in
lower volatility from a balance sheet perspective, ie, when liabilities
are taken into account.
Risk tolerance
Different companies have different risk tolerances and may not use
the same risk measurement metrics. Different measurements and
risk tolerances will lead to different optimal strategies. For example,
a company with a higher risk tolerance is likely to have a larger
allocation to commodities and inflation-sensitive equities, while a
company with a lower risk tolerance will gravitate towards TIPS
and floaters.
INFLATION RISK FOR LIFE INSURANCE COMPANIES
While most of the impact of inflation on P&C insurers profitability is tied to the adverse growth in claim payments, life insurance
companies are exposed in a different manner.
In fact, investment portfolios of life insurance companies are heavily weighted towards fixed-income investments such as government
and corporate bonds, and residential and commercial mortgages.
These investments typically pay fixed rates of interest and have
maturities much longer than those typically held by P&C companies.
As capital markets react to an inflation surprise, interest rates will
rise, decreasing the market value of these long-term fixed-income
investments. Depending on the timing of the investment purchases,
the unrealised loss position of the life insurers portfolio will grow,
leading ultimately to realised losses if and when the company needs
to sell assets.3
On a statutory accounting basis, realised losses are absorbed first
by the interest maintenance reserve (IMR) and then, once IMR is
depleted, by the companys capital, which will impair the insurers
ability to underwrite new business. On a Generally Accepted
Accounting Standards (GAAP) basis, losses will flow through to
386
i
i
i
perrucci 2012/7/11 17:12 page 387 #409
the company income statement, and both income and capital will
suffer.4 On an economic basis, the economic value of the company
will depend on the duration management of both sides of the balance sheet. If the duration of the assets is shorter than the duration
of the liabilities, the market value of the assets will fall less than the
market value of the liabilities such that the net surplus (assets minus
liabilities) position may actually benefit. However, if the reverse is
true and the asset duration is longer than the liabilities, the economic
value of the company may fall. Prudent assetliability management
is crucial in measuring and understanding this dynamic.
Beyond duration risk, negative convexity may also negatively
affect a life companys surplus position. Life companies typically
sell options on both sides of the balance sheet. For instance, on the
liability side, they give policyholders the right to lapse their policies
when interest rates rise and receive their policy cash values, which
may cause the company to sell assets just when market values have
been impaired. On the investment side, life companies often make
investments, such as mortgage-backed securities, that allow for the
return of funds to the company when interest rates fall, requiring
reinvestment in a lower yielding environment. The net effect is that
life companies typically suffer when there are large swings in interest
rates, regardless of the direction of the rate changes. Rapidly rising
interest rates in an increasing inflation environment are no exception.
Like P&C companies, life companies are exposed to higher
expenses during inflationary periods. Unlike P&C companies, which
may have some ability to re-price renewable policies during times of
increased inflationary pressure, life policies often require long-term
fixed contracts, which are priced using ex ante long-term inflation
projections. When actual inflation turns out to be higher than these
initial assumptions, life companies have a limited ability to recoup
the higher costs that result.
While inflation risk for life insurers can be significant, a mitigating
factor is that most life insurance benefits are not directly indexed to
inflation, to the contrary of what happens for P&C companies, who
must typically pay the full replacement cost of the insured item.
CONCLUSIONS
Given the monetary and economic environment at the time of writing, inflation risk is more relevant than ever. Simulation results
387
i
i
i
perrucci 2012/7/11 17:12 page 388 #410
INFLATION-SENSITIVE ASSETS
clearly show that the latter can indeed be the greatest financial risk
faced by P&C insurers, at least at long horizons (five years or more).
Asset classes that perform well during periods of higher inflation
are few, and those that exist have different characteristics and qualities. To lead to optimal enterprise performance, a successful strategy
must combine the knowledge required to invest in inflation-sensitive
assets, with the management experience and understanding of the
insurance business, and the needs of the specific client at hand.
All rights reserved. Protecting Insurance Portfolios from Inflation is licensed to Incisive Financial Publishing Ltd by Conning
Asset Management (Conning) and may not be reproduced or
disseminated in any form without the express permission of Conning. This publication is intended only to inform readers about
general developments of interest and does not constitute investment advice. While every effort has been made to ensure the accuracy of the information contained herein, Conning does not guarantee such accuracy and cannot be held liable for any errors in or
any reliance upon this information.
Conning does not guarantee that this publication is complete.
Opinions expressed herein are subject to change without notice.
Past performance is no indication of future results. Conning is
a portfolio company of the funds managed by Aquiline Capital
Partners LLC (Aquiline), a New York-based private equity firm.
1
However, as a counterargument, it is clear from Figure 16.4 that money velocity plummeted
during the same period.
For listed companies, stock price is a measure of economic value, although technical effects
can at times be important.
Clearly, what matters at the end is the duration difference between assets and liabilities.
388
i
i
i
perrucci 2012/7/11 17:12 page 389 #411
17
Markus Aakko
PIMCO
i
i
i
perrucci 2012/7/11 17:12 page 390 #412
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 391 #413
Indexation policy
US
UK
Netherlands
Canada
i
i
i
perrucci 2012/7/11 17:12 page 392 #414
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 393 #415
1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
exceeding 50%. Inflation can also feed further inflation, as indexation removes the ability of a reduction in real cost of production, postpones the necessary adjustments and makes the economy
susceptible to exogenous shocks (Durevall 1998).
Humankind has known inflation at least since records of prices
began. Often driven by local weather, prices of grain would fluctuate,
driving the cost of living up and down, as costs associated with
transportation did not allow for substantial trade between areas with
food surpluses and deficits (see, for example, Fischer 1996, pp. 923).
Another strong influence on price levels came from new discoveries
of precious metals, used as money through the expansion of money
supply (Pettee 1936).
The modern period2 saw the establishment of central banks
and monetary policy, and a shift in inflationary tendencies: weather
plays a diminished role due to robust logistical networks around the
globe, and the main drivers of inflation are now supply and demand
of key commodities, labour costs and the activities of central banks.
During the 20th century, the key events that shaped public consciousness of inflation were the hyperinflationary period in Germany (1923), which gave the German Central Bank (the Bundesbank,
established in 1957) and its effective successor, the European Central
Bank (ECB), a permanently hawkish tone, the deflationary period
during the Great Depression in the US (1930s) and the inflationary
decade of the 1970s in most Western countries.
393
i
i
i
perrucci 2012/7/11 17:12 page 394 #416
INFLATION-SENSITIVE ASSETS
ial assumptions.
As mentioned earlier, pre-funding liabilities mitigates the risk of
whether of not the company will actually meet its pension obligations in the future (in other words, the counterparty risk of
394
i
i
i
perrucci 2012/7/11 17:12 page 395 #417
i
i
i
USA
2,121
Corporate AA
Corporate AAAA
bonds
Usually 7 yr
UK
1,485
Corporate AA
Trustees select
discount rate basis
Usually up to 10 yr
PPF
Euro-swap curve
Usually up to 15 yr
None
Commuted value
assumptions,
depending on
liabilities
None
Netherlands
Canada
Ontario
Discount
curve
(accounting)
1,079 High-quality
corporate bond rate
(IAS 19)
845
Yield on high-quality
fixed income
Discount
curve
(funding)
Deficit
amortisation
National
insurance
PBGC
has one for the Province. PBGC, Pension Benefit Guarantee Corporation. PPF, Pension Protection Fund. Accounting curve: the discount
rate used to determine the present value of liabilities in the financial statements of the sponsor. Funding curve: the discount rate used to determine
the present value of liabilities for assessing the required annual contribution under local pension regulation. Deficit amortisation: the period defined
by local pension regulation as the maximum allowed for calculating the required annual contribution. For example, a seven-year amortisation means
that any deficit must be funded in installments over seven years or less. National insurance: collective, typically government-sponsored, funds that
act as insurance for participants against insolvency of pension plans.
Sources: Halonen (2011), data as of December 31, 2010; Pensions Protection Fund (2010); OECD (2007).
Country
Private
pension
assets
(US$ bn)
INFLATION-SENSITIVE ASSETS
396
Table 17.2 Private pension plans discount curve and other characteristics
i
i
i
perrucci 2012/7/11 17:12 page 397 #419
(US$ million)
Liabilities
(US$ million)
Equities
Fixed income
216
540
Present value
of liabilities
1,000
Alternatives
144
Total assets
900
Total liabilities
1,000
rate. These are also the most easily observable changes in real time,
as actuarial calculations are usually done infrequently,7 and inflation
indexes are typically released with a lag of several weeks or months.
MEASURING THE INTEREST RATE SENSITIVITY OF
LIABILITIES
Most plans measure their sensitivity to changes in discount rates
using interest rate duration. By calculating duration on both assets
and liabilities, we can use the difference (the duration gap) to estimate changes in the surplus based on various market scenarios. Most
plans are underweight in duration, which is to say that their liabilities have a substantially higher sensitivity to changes in interest rates
than their assets do.
As an illustration, consider the example of a hypothetical US pension plan that is 90% funded ($900/$1,000), and has 60% of liabilities
whose cashflows are linked to CPI.8 The plan has also taken significant steps towards reduction of surplus volatility through duration
overlay strategies. The characteristics of this pension plan are shown
in Table 17.3. The plan employs an interest rate swap overlay for a
notional amount of US$540 million (60% of assets) with duration of
8.7 years. Due to a recent merger of two corporate plans (one with full
indexation and one entirely nominal), 60% of pension liabilities are
indexed to inflation through a COLA. The overall liability duration
is 12.7 years. The resulting balance sheet is shown in Table 17.4.
The duration of assets is the sum of the weighted duration of equities and fixed income securities, that is 8.2 years, while liabilities have
a weighted duration of 14.1 years, leaving a negative duration gap
of 5.9 years. In other words, a rise in interest rates is beneficial, as
it decreases the amount by which the pension fund is underfunded.
397
i
i
i
perrucci 2012/7/11 17:12 page 398 #420
INFLATION-SENSITIVE ASSETS
Weight
(% of assets)
Equities
Fixed income
Overlay
Alternatives
Liabilities
Duration
(years)
24
60
60
16
111
N/A
5
8.7
N/A
12.7
Weighted
duration
(years)
3.0
5.2
14.1
5.9
Total
(17.1)
i
i
i
perrucci 2012/7/11 17:12 page 399 #421
Weight (%)
Real rate duration (years)
Inflation duration (years)
Assets
Liabilities
100
8.2
8.2
111
12.7
5.1
Surplus
5.9
2.6
That is 40% 12.7 = 5.1 years, reflecting the liabilities not linked to
inflation. We assume that there are no inflation-linked instruments among
the assets.
The implications of contribution requirements for the plan are further complicated, as liabilities are marked-to-market using nominal
discount rates, typically on a monthly basis, while a rise in inflation
expectations would be reflected through actuarial cash projections
on a less frequent basis (typically a few times per year). In addition
to delays, actuarial assumptions often include further lags due to
smoothing assumptions.
Using cashflow (inflation) duration also assumes that there is the
same change in inflation expectations for all terms across the yield
curve (this parallel shift assumption is common to all duration
measures). Finally, meaningful changes in inflation and interest rates
will typically not happen without some attendant changes in the valuation of other asset classes. As a result of the complex interrelation
of equity returns, spreads, interest rates and inflation, the duration
measure is a useful, but by no means comprehensive, measure of
risk.
RISK BUDGETING
Sensitivity measures such as duration are useful in evaluating
expected changes in surplus given changes in a single market variable. An alternative, and more comprehensive, measure is volatility or, in the case of pension funds, volatility of the pension surplus. To estimate surplus volatility, most plans use a covariance
matrix, derived from historical data, forward-looking views or some
combination of the two.
By obtaining an estimate for both surplus volatility and required
return on surplus, we can set up an optimisation problem. Specifically, using modern portfolio theory, it is possible to seek optimal
allocations of assets in order to minimise surplus volatility, under
399
i
i
i
perrucci 2012/7/11 17:12 page 400 #422
INFLATION-SENSITIVE ASSETS
Asset class
Equity
Fixed income
Alternatives
Liabilities
Overlay
Wt
(%)
Vol.
(%)
S&P 500
24
BarCap US Aggregate
60
DJ/CSFB Hedge Funds Index
16
BarCap US Long Corporate A and higher 111
BarCap 10-year Swap Bellwether
60
16
4
8
9
8
Proxy index
the constraint of achieving the return required for the plan. The
resulting optimal allocation strategy is usually called a risk budget. As inferred by the name, a risk budget is meant to help pension
plan managers allocate a scarce resource, surplus risk, across various investments. A typical way of analysing the risk budget is risk
decomposition, which is the process of estimating the contribution
to surplus variance arising from each investment. If denotes the
row vector of weights for each asset in the portfolio, and denotes
the covariance matrix, we can calculate the overall variance of the
surplus (Litterman et al 2003) as
2 = T
(the superscript T indicates transposition). We can then decompose the overall variance into each individual assets contributions.
Letting i denote the ith row in the covariance matrix and 2 denote
the portfolio variance, the percentage contribution to the overall
variance for asset i is
i i
T
2
(17.2)
i
i
i
perrucci 2012/7/11 17:12 page 401 #423
Fixed income
Alternatives
Liabilities
0.1
1.0
0.2
0.9
0.9
0.6
0.2
1.0
0.2
0.0
0.2
0.9
0.2
1.0
0.7
1.0
0.1
0.6
0.2
0.1
Overlay
0.1
0.9
0.0
0.7
1.0
Weight (%)
Equities
Fixed income
Overlay
Alternatives
Liabilities
Variance
decomposition (%)
24
60
60
16
111
25
11
13
4
95
Total
100
Return-seeking
assets
Variance
contribution
(%)
Liabilities
and hedges
Equities
Alternatives
25
4
Liabilities
Fixed income
Overlay
Total
29
Total
Variance
contribution
(%)
95
11
13
71
Given that the liability proxy consists of corporate bonds, positive correlation between equities and corporate bonds reduces the contribution
to risk attributable to equities. If a Treasury index had been used to proxy
liabilities, equities and alternatives would have contributed a further 48%
to total surplus variance.
(ii) The fixed income assets are invested in the Barclays Capital
US Aggregate Index, which has a duration of five years.10
(iii) The interest rate overlay can be modelled using the Barclays
Capital 10-year Swap Bellwether Index.
401
i
i
i
perrucci 2012/7/11 17:12 page 402 #424
INFLATION-SENSITIVE ASSETS
By using the appropriate covariance matrix (Table 17.7), we can calculate the total surplus volatility, expressed as a percentage of assets
or in US dollar terms. For our example, we obtain an overall volatility
= 6%, or about US$54 million (the assets value is US$900 million).
The next step is to decompose the overall portfolio variance into
its individual contributions, based on Equation 17.2 (Table 17.8). As
it is clear from this risk decomposition, the largest influence on the
total surplus variance comes from the liabilities (Table 17.9). Negative contributions to risk, from the fixed income sector (11%) and
the interest rate overlay (13%), are a result of their being hedging
assets, or, in other words, assets highly correlated with the liabilities,
thus reducing overall volatility.
It is useful to consider each asset and investment as either a returnseeking asset or a liability-hedging asset. Return-seeking assets
(equities and alternatives in our example), which typically do not
hedge liabilities, provide an opportunity for additional returns in
exchange for higher risk. Conversely, liability-hedging assets (fixed
income and the interest rate overlay in our example) reduce surplus
volatility, but their expected return might be less attractive compared
with the liability discount rate.
With this type of analysis, risk managers can refine their allocation process, reduce unwanted risks and optimally allocate their
risk budget and capital to asset classes, in line with the goals of the
pension fund. For underfunded plans, this may mean increasing
allocation to return-seeking assets, at the expense of hedging assets,
thus increasing the overall surplus volatility. For fully funded plans,
especially the ones that are closed to new entrants, the plans objectives may be best served by closely matching assets and liabilities,
and by minimising surplus volatility.
FROM ASSET-CLASS MODELS TO FACTOR MODELS
Each of the asset classes considered in the risk budgeting analysis
above is sensitive to several risk factors, including interest rates,
inflation and equity. Some sectors embody one risk factor in particular (eg, US Treasuries have interest rate and inflation risk; common
shares mainly have equity risk), while others present a more complex combination of sensitivities (eg, convertible bonds are sensitive to interest rates, inflation, volatility, equity prices and corporate
spreads).
402
i
i
i
perrucci 2012/7/11 17:12 page 403 #425
i
i
i
perrucci 2012/7/11 17:12 page 404 #426
INFLATION-SENSITIVE ASSETS
0.0
0.0
0.0
1.0
R2
N/A
N/A
N/A
N/A
N/A
Fixed income
Coefficients t -stat.
Intercept
Duration risk
Corporate risk
Equity risk (S&P 500)
0.0
3.8
0.3
N/A
R2
0.9
10.6
56.5
16.8
N/A
Alternatives
Coefficients t -stat.
Intercept
Duration risk
Corporate risk
Equity risk (S&P 500)
0.0
0.9
0.5
0.2
R2
0.4
4.0
1.9
4.0
6.8
Liabilities
Coefficients t -stat.
Intercept
Duration risk
Corporate risk
Equity risk (S&P 500)
0.0
7.9
1.1
N/A
R2
0.9
0.6
33.3
21.0
N/A
Since both the equity asset class and the equity risk factor are defined by
the S&P 500, the factor loading is 1.0. Duration risk is expressed as unit
of volatility per year of duration. Hence, the coefficient value approximates
the duration of the asset.
404
i
i
i
perrucci 2012/7/11 17:12 page 405 #427
Table 17.11 Risk factor loadings for the pension funds surplus
Assets and
liabilities
Equities
Fixed income
Overlay
Alternatives
Liabilities
Surplus
Weight (%)
Equity
24
60
60
16
111
11
1.0
0.0
0.0
0.2
0.0
0.3
Duration
0.0
3.8
8.7
0.9
7.9
1.2
Corporate
0.0
0.3
0.0
0.5
1.1
1.0
Duration
0.1
1.0
0.3
Corporate
0.5
0.3
1.0
in Table 17.10. Note that the intercepts are, in most cases, statistically indistinguishable from zero. The swap overlay is not shown in
Table 17.10, since, by virtue of our choice of factors, it will have a
non-zero loading to duration risk only, as displayed in Table 17.11.
Based on the risk factor loadings and the weights of assets and liabilities, we can assess the factor exposures of the pension fund surplus
(Table 17.11).
After deriving the covariance matrix for the three factors (Table
17.12), we calculate the overall volatility of the surplus,11 as well as
the variance decomposition and risk contribution from each factor
(Table 17.13). The variance decomposition by factor offers a unified
approach to examining the underlying risks within each asset class
as well as in the overall portfolio or pension fund surplus.
Most notably, the corporate spread risk factor now plays a prominent role, highlighting the mismatch between the asset side (where
the Barclays Capital Aggregate Index used as a proxy contains limited corporate exposure) and the liability side, where the large factor
loading arises from the corporate curve used for discounting. As a
result, the surplus benefits from a widening in credit spreads. This
is also the reason why the duration loading of the surplus at 1.2 is
405
i
i
i
perrucci 2012/7/11 17:12 page 406 #428
INFLATION-SENSITIVE ASSETS
Risk factor
Equity
Duration
Corporate
Total
Factor-model surplus volatility
Loading
0.3
1.2
1.0
Variance
decomposition (%)
47
3
50
100
4
less in absolute value than that found when analysing risk by asset
class, as the negative duration is now split between the (swap-like)
duration factor and the corporate risk factor.
Further insight comes from comparing the correlation structures
of asset classes versus risk factors. While equities and liabilities
were previously positively correlated (Table 17.7), risk factor analysis breaks this (as seen in Table 17.12) into a negative correlation
between equity risk and interest rate duration (ie, falling equity
prices often coincide with rising swap prices) and a positive correlation between equity returns and corporate bonds excess returns (ie,
equities drop when corporate bond prices fall in relation to treasury
bonds).
Risk factor analysis can provide considerable insight on the risk
and return drivers of any portfolio. As usual, the selection of factors
and the empirical data (or implicit views) used for estimation will
have an important effect on the covariance matrix and model outputs
and thus need to be scrutinised in detail.
MODELLING INFLATION IN A FACTOR FRAMEWORK
The factor framework is useful for modelling risk factors that influence the performance of several asset classes. Inflation is an example of such a factor. Factor or asset-class volatility models typically
use nominal returns and volatilities as inputs. Inflation is therefore
represented as a static expectation contributing to asset-class returns.
One approach to addressing inflation explicitly is to move from
the nominal to the real return space, which would include deriving
406
i
i
i
perrucci 2012/7/11 17:12 page 407 #429
0.3
1.5
2.2
129
an inflation-adjusted covariance matrix. This may pose some computational issues, as inflation data is published with a lag and less
frequently than market prices, which can be observed in real time.
Moreover, this may limit the ability to rely on third-party vendors for
market return data series, as these are almost invariably calculated
in nominal terms.
Another way to go about this is to include a new variable: unanticipated or excess inflation, which would represent inflation in excess of
any actuarial assumptions. According to the Fisher equation (Equation 17.1), most asset classes embed some level of expected inflation iEXP in their price; in other words, their market value is derived
by discounting future cashflows using a nominal curve. Since both
actuarial projections (used for cashflow projections) and nominal
discount curves already include an inflation assumption, the only
remaining issue for pension plans is the effect of excess inflation over
the relevant (typically long-term) time horizon. In practice, things
are not so simple, as market prices (ie, discount curves) adjust in
real time, while the actuarial assumptions used to estimate future
liabilities cashflows do not.
With the advent of the inflation-linked bond market, we have
a market-based measure of expected inflation (for different tenors
across the term structure), also known as break-even inflation.
For example, the 10-year US Treasury Inflation-Protected Securities (TIPS) break-even, a measure of long-term inflation expectations, has ranged from 0.1% to 2.7% (Figure 17.2). One drawback of
these market-based measures is the short history of the data series,
which, in particular, does not contain any period of high inflation
(US TIPS were first issued in 1997). In addition, there is evidence
of a substantial liquidity premium affecting break-even inflation in
407
i
i
i
perrucci 2012/7/11 17:12 page 408 #430
INFLATION-SENSITIVE ASSETS
Jan 4, 2011
Jan 4, 2010
Jan 4, 2009
Jan 4, 2008
Jan 4, 2007
Jan 4, 2006
Jan 4, 2005
Jan 4, 2004
Jan 4, 2003
Jan 4, 2002
Jan 4, 2001
Jan 4, 2000
Jan 4, 1999
0.5
Source: Bloomberg.
the US, at least for the first few years since first issuance (because
of this, inflation break-evens are likely to underestimate inflation
expectations in at least part of the period considered).
It is possible to extend these market-based time series back in time
for more meaningful historical comparisons, although the methods used to do this will introduce their own bias. One option is to
use information from publicly available inflation surveys (although
these surveys typically poll expectations over a short time horizon:
six months in the following analysis).
In Tables 17.14 and 17.15, we look at the excess inflation series
derived as the difference between actual realised inflation rates and
inflation expectations. The latter are measured by the median of
the Livingston Survey (specifically, the median of the semi-annual
survey is used as an estimate of inflation expectations for the next
six months).12
It should be noted that the data shows two very different periods:
during 1947 to 1974, the surveyed economists persistently underestimated inflation (on average by 0.5%), whereas between 1975 and
2011, economists persistently overestimated inflation (on average
by 0.9%).
408
i
i
i
perrucci 2012/7/11 17:12 page 409 #431
19471974
19752011
Mean (%)
Standard deviation (%)
0.5
1.5
0.9
t-statistic
2.3
6.0
Observations
57
72
1.3
16
1
4
1
0.1
0.5
0.1
0.1
0.4
1
0.3
0.5
0.3
0.1
1
0.2
0.2
1
0.4
i
i
i
perrucci 2012/7/11 17:12 page 410 #432
INFLATION-SENSITIVE ASSETS
Loading
Equity
Duration
Corporate
Excess inflation
Variance
decomposition (%)
0.3
0.8
1.5
0.8
42
1
46
13
Total
100
Regressing the excess inflation series against changes in 10year nominal Treasury yields (as a measure of long-term inflation
expectations),14 we find that a 1% inflation surprise moves the 10year nominal yield by 10 basis points. In other words, the excess
inflation factor loading (as a percentage of assets) is equal to the
product of
the regression coefficient (+0.1) linking short-term survey
In detail
excess inflation factor loading = 111% 0.1 60% 12.7
= 0.8
i
i
i
perrucci 2012/7/11 17:12 page 411 #433
One of the challenges for the inflation market has always been
that few natural sellers of inflation-hedging instruments exist, while
demand for inflation assets is definitely growing. In Mercer Consulting (2011), 80% of pension plans in Europe view inflation as a
major concern. The predominant investment vehicles for these plans
are inflation-linked bonds (18% of respondents), followed by other
investments, such as commodities and real estate (12%). European
plans on average hold 39% in real estate, depending on the size
of the plan. In comparison, the average plan has only 1.8% in commodities, with only 7.5% of European funds investing in this asset
class.
411
i
i
i
perrucci 2012/7/11 17:12 page 412 #434
INFLATION-SENSITIVE ASSETS
Country
Outstanding notional
(US$ bn)
Issuance in 2010
(US$ bn)
213
146
62
21
568
260
55
28
30
28
21
15
0
92
50
0
1
2
1,384
209
France
Italy
Germany
Greece
US
UK
Japan
Sweden
Canada
Total
i
i
i
perrucci 2012/7/11 17:12 page 413 #435
pension funds; state and other public pension funds have been the
largest pension investors, given their direct indexation of liabilities.
In continental Europe, one of the potentially largest markets is the
Dutch pension market, given that their existing inflation-linked liabilities remain largely unhedged. In the UK, where inflation-linked
bonds are well established, approximately one-third of the existing
stock is held by pension funds, and another third is held by insurance companies (Barclays Capital 2010). Trends in pension buyouts
in the early 2000s have also increased demand, as buyers of liabilities
seek to hedge their exposures.
Inflation derivatives
Inflation derivatives can offer more tailored protection against inflation. The market has grown rapidly, especially in Europe, and has
matured over time. The inflation derivatives market can be divided
into three basic instruments:
1. inflation futures;
2. inflation swaps;
3. inflation options (such as caps and floors).
Futures markets in inflation have struggled around the globe. In
the US, the most recent attempt was the CPI-linked futures contract
launched in 2004, which subsequently ceased to trade given the lack
of demand.15 The European Harmonised Index of Consumer Prices
(HICP) futures are traded at both the Chicago Mercantile Exchange
(CME) and the Eurex, but volumes have been virtually non-existent
between 2009 and the time of writing.
The swap and option markets, on the other hand, have flourished.
In the US, there has been demand for inflation total-return swaps,
and option volumes have been rising substantially (Figure 17.3).
Because of their customisable nature, inflation derivatives probably
hold the greatest promise for the pension industry.
Option strategies could also turn out to be the most efficient way
to hedge against tail risks in inflation. Traditional tail-risk hedging strategies such as out-of-the-money equity puts, collars and
put-spread collars can now be implemented using inflation options
instead. It should be noted that some of these strategies can get
very complex and, given the relatively recent introduction of inflation derivatives, for most institutional investors they may require a
reliance on outside expertise.
413
i
i
i
perrucci 2012/7/11 17:12 page 414 #436
INFLATION-SENSITIVE ASSETS
7,225
7
6
5
4
3
2
1,365
1
0
460
2008
2009
2010
i
i
i
perrucci 2012/7/11 17:12 page 415 #437
CaseSchiller
CPI-U
20
15
10
5
% 0
5
10
15
25
Jan 1988
Jan 1989
Jan 1990
Jan 1991
Jan 1992
Jan 1993
Jan 1994
Jan 1995
Jan 1996
Jan 1997
Jan 1998
Jan 1999
Jan 2000
Jan 2001
Jan 2002
Jan 2003
Jan 2004
Jan 2005
Jan 2006
Jan 2007
Jan 2008
Jan 2009
Jan 2010
20
i
i
i
perrucci 2012/7/11 17:12 page 416 #438
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 417 #439
i
i
i
perrucci 2012/7/11 17:12 page 418 #440
INFLATION-SENSITIVE ASSETS
CPI-U
40
30
20
10
Dec 1, 1969
Jun 1, 1970
Dec 1, 1970
Jun 1, 1971
Dec 1, 1971
Jun 1, 1972
Dec 1, 1972
Jun 1, 1973
Dec 1, 1973
Jun 1, 1974
Dec 1, 1974
Jun 1, 1975
Dec 1, 1975
Jun 1, 1976
Dec 1, 1976
Jun 1, 1977
Dec 1, 1977
Jun 1, 1978
Dec 1, 1978
Jun 1, 1979
Dec 1, 1979
Source: Bloomberg.
only put pressure on earnings but also create uncertainties, and thus
increase the equity risk premium, setting the stage for poor equity
performance.
However, it should be noted that, while equities as an asset
class are not an effective inflation hedge, there may be specific
equity sectors with better inflation-hedging properties. Examples
could include commodity producers, such as gold mines, or essential utilities, such as sewer and electricity generators. Investors
could conceivably create baskets of equity securities in multiple
inflation-sensitive industries, in order to reduce business cycle
effects.
CONCLUSIONS
Whether inflation is a material risk for a pension plan is largely determined by its specific circumstances. The hedging of nominal liabilities, which has been a substantial trend in the industry since the mid
2000s, has relied on the easily identifiable duration gap between
assets and liabilities. Analysing inflation sensitivity of the plans in
the same way is much more difficult, given the complexities associated with both actuarial assumptions and the estimation of the
inflation sensitivity of the underlying investments.
418
i
i
i
perrucci 2012/7/11 17:12 page 419 #441
Considering hedges for inflation risk makes sense, but, given the
intrinsic uncertainty associated with estimating the sensitivity to
inflation, a partial hedge might be a more practical trade-off. Furthermore, the hedge instruments directly linked to inflation, such
as inflation-linked bonds or derivatives, can often be costly, as there
are few natural sellers of inflation protection. For others, basis risk
can be substantial. The latter, however, can be mitigated by diversification, ie, by employing several of the inflation-hedging strategies
at once as a way of reducing the idiosyncratic risk arising from each
asset class.
Lastly, pension plans should be concerned about inflation tail risk.
While the implications of moderate inflation are not dire to either
the plan sponsors or the pension beneficiaries, substantial and protracted inflation, similar to that experienced in the 1970s, could have
a large negative influence on pension funds solvency ratios and their
ability to meet their obligations.
1
The plan sponsor is the entity establishing the pension programme. It might be a State or local
government entity, or a private corporation.
Effectively the 20th century, although some central banks did come into existence earlier than
this.
Surplus is equal to the market value of assets minus the market value of liabilities.
For example, the US Generally Accepted Accounting Principles (GAAP) and the International
Accounting Standards (IAS).
See, for example, the discussion on UK public pension discount practices in Ralfe (2011).
For a detailed study on relationship of wage growth and price inflation, see, for example,
Hess and Schweitzer (2000).
While the pension plan data used here for illustration is based on an actual plan, we have
made many simplifying assumptions. In the actual case, asset classes are represented with a
higher degree of granularity and the relationship between the plans liability and inflation is
substantially more complex. We have used proxy benchmarks to represent both assets and
liabilities in the illustrative calculations. The size of the pension fund liability is changed to
US$1 billion for illustrative purposes.
Of course, we might pick a different pair of variables (nominal rates and inflation, or nominal
rates and real rates). For example, in the section discussing the modelling of inflation in a factor
framework (page 406), we shall work with nominal rate duration (swap rates discounting)
and inflation.
10 Duration estimates for the indexes are based on information from Barclays Capital as of
March 31, 2011.
11 This volatility turns out to equal 4% of the asset value per annum, or about US$36 million
per year. It should be noted that, given the more granular approach, the liability risk appears
slightly smaller than that using the asset-class-based approach. This is an illustration of how
different covariance matrices may lead to shifts in relative contributions, and overall risk, in
different models.
12 See http://www.philadelphiafed.org/ for details on this inflation survey.
419
i
i
i
perrucci 2012/7/11 17:12 page 420 #442
INFLATION-SENSITIVE ASSETS
13 Data used to estimate factor correlation and volatility for the equity, duration and corporate
spread factors represents a shorter period than that used for excess inflation. We have used
monthly data from January 1994 to the end of December 2010 for the overall factor model,
and augmented it with excess inflation using semi-annual data from January 1972 to the end
of December 2010.
14 We could have used the 10-year US TIPS inflation break-even instead. We elected to use the
nominal Treasury yield in order to avoid the issues with limited data and technical liquidity
distortions in the US TIPS market.
15 Previous attempts to establish the inflation futures include, for example, the CPI futures
launched in 1985 by New Yorks Coffee, Sugar and Cocoa Exchange.
16 See OECD Global Pension Statistics data at http://www.oecd.org/daf/pensions/gps.
REFERENCES
420
i
i
i
perrucci 2012/7/11 17:12 page 421 #443
Hess, G. D., and M. E. Schweitzer, 2000, Does Wage Inflation Cause Price Inflation?,
Federal Reserve Board of Cleveland, Policy Discussion Paper 1, April.
Kutsch, N., and C. Lizieri, 2005, The UK Pension Industry: A Research Report for the
Pension Real Estate Association, University of Reading Business School, December.
Litterman, B., and The Quantitative Resources Group, 2003, Modern Investment Management: An Equilibrium Approach (Hoboken, NJ: John Wiley & Sons).
Mercer Consulting, 2011, Asset Allocation Survey: European Institutional Market Place
Overview 2011, Report, May.
Mitchell, O., 2000, New Trends in Pension Benefit and Retirement Provisions, Pension
Research Council Working Paper (PRC WP 2000-1).
NAREIT, 2011, REITWatch: A Monthly Statistical Report on the Real Estate Investment
Trust Industry, National Association of Real Estate Investment Trusts, April.
Nazmi, N., 1996, Economic Policy and Stabilization in Latin America (New York: M. E. Sharpe).
OECD, 2007, Global Pension Statistics 2007, URL: http://www.oecd.org.
OECD, 2011, Pensions at a Glance 2011: Retirement-Income Systems in OECD 180 and
G20 Countries, Report.
OECD Secretariat, 2009, Survey of Investment Regulation of Pension Funds, URL:
http://www.oecd.org/dataoecd/30/34/2401405.pdf.
Pensions Protection Fund, 2010, The Purple Book 2010, The Pensions Regulator and
Pensions Protection Fund.
Pettee, E. W., 1936, Long-Term Commodity Price Forecasting 1850 to 1930, The Journal
of Business of the University of Chicago 9(2), p. 97.
Piggott, J., and R. Sane, 2009, Indexing Pensions, Report, The World Bank.
Ralfe, J., 2011, The Correct Pension Discount Rate, Financial Times, March 13.
Reserve Bank of Australia, 2007, The Recent Rise in Commodity Prices: A Long-Run
Perspective, Reserve Bank of Australia Bulletin, April.
Rockoff, H., 1984, Drastic Measures: History of Wage and Price Controls in the United States
(Cambridge University Press).
Schmitt, D. G., 1984, Postretirement Increases under Private Pension Plans, Bureau of
Labor Statistics 107(9), p. 3.
Scitovsky, T., 1979, Home Truths about Inflation, in Essays in Post-Keynesian Inflation,
pp. 2830 (Cambride, MA: Ballinger).
Shlaes, A., 2007, The Forgotten Man: A New History of the Great Depression (New York:
HarperCollins).
Siegel, L., and M. Waring, 2004, TIPS, the Dual Duration, and the Pension Plan, Financial
Analysts Journal 60(5), pp. 5264.
Standard & Poors, 2011, Commodities 101: Understanding Commodities and S&P GSCI, S&P
Indexes Practice Essentials Series, Standard & Poors Research.
Taborsky, M., and S. Page, 2010, The Myth of Diversification: Risk Factors vs Asset
Classes, PIMCO Viewpoints, September.
Towers Watson, 2010, Global Alternatives Survey, Towers Watson and the Financial
Times, June.
421
i
i
i
perrucci 2012/7/11 17:12 page 422 #444
INFLATION-SENSITIVE ASSETS
Towers Watson, 2011, Global Pension Asset Study 2011, Towers Watson and the Financial
Times, February.
Weinstein, H., 1997, Post-Retirement Pension Increases, in Compensation and Working
Conditions, Bureau of Labor Statistics, Fall, p. 49.
Yermo, J., 2007, Reforming the Valuation and Funding of Pension Promises: Are Occupational Pension Plans Safer?, OECD Working Papers on Insurance and Private Pensions,
No. 13.
422
i
i
i
perrucci 2012/7/11 17:12 page 423 #445
18
i
i
i
perrucci 2012/7/11 17:12 page 424 #446
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 425 #447
i
i
i
perrucci 2012/7/11 17:12 page 426 #448
INFLATION-SENSITIVE ASSETS
US CPI
250
200
150
100
50
0
1961 1966 1971 1976 1981 1986 1991 1996 2001 2006
Sources: HEPI, July 1June 30 data (Research Associates of Washington and
Commonfund Institute). CPI data (US Department of Labor) is calculated to July 1
June 30 (annual published CPI is computed over the calendar 12-month period).
Based at 100 in 1983.
Sep
2001
Mar
2003
Sep
2004
Mar
2006
Sep
2007
Mar
2009
Sep
2010
i
i
i
perrucci 2012/7/11 17:12 page 427 #449
Oct
2002
Jan
2004
Apr
2005
US CPI
Jul
2006
Oct
2007
Jan
2009
Apr
2010
600
400
200
0
Jan
1985
Apr
1988
Jul
1991
Oct
1994
Jan
1998
Apr
2001
Jul
2004
Oct
2007
Jan
2011
i
i
i
perrucci 2012/7/11 17:12 page 428 #450
INFLATION-SENSITIVE ASSETS
REAL ASSETS
The perceived realities of UHNW inflation are a simple explanation
why consumer-inflation-linked securities do not provide an inflation solution. With the low real yields that characterised the 2008
global financial crisis and are still predominant at the time of writing in 2011,1 such an investor is, in practice, bearing negative real
returns. Additionally, wealthy investors have an absolute return target profile, ie, an asymmetrical return target of inflation or more.
In comparison, investment institutions have traditionally tended to
have a more symmetrical benchmark approach. Behavioural finance
has shown if, indeed, it was ever in doubt that investors have
a natural bias against loss: they find losses more painful than gains
pleasurable. Since UHNW investors are less intermediated than the
beneficiaries of institutions, their preferences carry through more
clearly to the market. Of course, the fact that absolute returns are
desired does not mean that they are readily available, but UHNW
investors themselves provide the backdrop and the capital for the
financing and generation of new opportunities for absolute return
investments.
It is exactly the wish to own inflation-protected assets with reasonably high returns that sheds light on the historically material allocation to real assets within UHNW portfolios: particularly real estate,
but also agriculture and forestry.2 Good-quality hard assets, in the
right location, with stable inflation-exposed cashflows (be they core
real estate, infrastructure, power plants or even forestry) are attractive for many investment portfolios. Historically, the 69% nominal
returns on these investments have translated into real, unleveraged
returns in the 46% range (once adjusted by consumer inflation). But
even for real assets, returns have decreased over time, partly because
of increasing institutional investors demand. Hence, two strategies
have been employed by UHNW investors in order to retain inflation
protection while generating target returns: developing real assets
from an early stage and using leverage.
Part, of course, of the return available from very early-stage investment is a premium for illiquidity, which wealthy investors have
historically been willing to receive. Their long, inter-generational,
time horizon makes them natural buy-and-hold or buy-and-wait
investors.
428
i
i
i
perrucci 2012/7/11 17:12 page 429 #451
i
i
i
perrucci 2012/7/11 17:12 page 430 #452
INFLATION-SENSITIVE ASSETS
Construction
Commissioning
Operation
operating real asset. The stage immediately prior to operation is commissioning: letting a completed, or soon-to-be completed, property
or switching on a wind farm. Forestry fits less tidily, but commercial
thinning (getting the forest ready for harvest) is analogous. Construction or planting precedes and typically represents the largest
single item of cost to create the asset. Construction of real estate,
erection of wind turbines, building roads and planting trees are all
capital-intensive activities.
Project development can be hugely value accretive, and commensurately risky. At the outset, it involves having the idea to create a real
asset, ie, producing a winning concept, and then securing the three
necessary ingredients for successful development. The first one is to
secure use of the land. At the outset, a developer might take an option
but, in the end, before beginning construction, will want to have the
absolute right to use the land in the way they need. Next are plans
and permissions: a building, or likewise an energy-producing asset,
needs detailed plans from concept design to detailed construction
drawings. And, in most cases, public authority permissions are also
required. The final element of the triumvirate is finance. Figure 18.5
gives a conceptual illustration of this value chain.
RISKS AND RETURNS
At the final stages of the value chain, the owner largely bears the
market risk of the asset, with the expectation that there is inflation
protection. Take the example of real estate. If an investor has developed a core office property, they will let it into the market for office
space. There is some idiosyncrasy around building quality and location, but its impact will be relatively small in relation to the prevailing rents in the area. In the case of real assets that produce fungible
commoditised outputs, such as energy or timber, there is scope for
trading, but the price is very largely market determined.
As you move to the beginning of the value chain, the risk is more
idiosyncratic, and the expected return is generally higher. Taking the
430
i
i
i
perrucci 2012/7/11 17:12 page 431 #453
three basic building blocks, the risks offer different types of return
at each stage. Most isolated from broader macro factors is the design
process. A clever, beautiful or efficient design is a purely idiosyncratic (ie, not correlated to market factors) source of return. This is
evident in real estate projects, as well as in decisions about the layout
of a timberland plantation, a wind farm, or solar park.
As for acquiring land, development land is most often priced by
residual value calculations. That is, you determine what the final
product, such as an office building, will be worth; you then deduct
costs, including financing, and a profit for the developer, usually
expressed as a percentage of costs, before arriving at the residual
value for the land. This leaves the value of land exposed to the current assessment of discounted future value of the end product. Of
course, this exposure to macro risks embeds some inflation sensitivity: if the price of real estate rises, land values will increase to reflect
this. And, as noted above, it exposes the residual value of the land
to the availability of finance: low discount rates both reduce financing costs and increase the discounted future value of the completed
asset. Consequently, the price of development land can be wildly
volatile through the cycle.
When it comes to permits, public authority permissions are largely
independent of market risk. In fact, public authority decisions relate
to policy and the wishes of their electorate, be it national, regional
or municipal. These should be largely uncorrelated to broader economic factors. Indeed, achieving preliminary, and then final, public
authority permits is highly value accretive, but also risky in a relatively binary fashion. A refusal to approve a new shopping centre is
hard to mitigate by means other than simply appealing. The extent
to which this development stage is dependent on broader macro factors lies in the possibility that, because of hard economic times, for
example, a public authority might restrict a building permit in order
to avoid undermining existing asset values.
Finally, the availability of finance is most closely correlated to
broader macro factors. In times of capital plenty, more projects are
financeable. In harder times, securing capital might generate more
return. Experience suggests that the benefits of capital plenty are
only fleetingly captured by developers and that after an interval the
surplus thus created seeps to other parts of the industry, often the
landowners.
431
i
i
i
perrucci 2012/7/11 17:12 page 432 #454
INFLATION-SENSITIVE ASSETS
Return (%)
Figure 18.6 Risks and returns along the real asset value chain
25
10
8
6
Development risk
Construction risk
Commissioning risk
Inflation and beta
Financed
and permitted
Construction
completion
Operational
start-up
i
i
i
perrucci 2012/7/11 17:12 page 433 #455
Return (%)
25
Development risk
+5%
10 Construction risk
8 Commissioning risk
6
+2%
+1%
Construction
completion
Operational
start-up
i
i
i
perrucci 2012/7/11 17:12 page 434 #456
INFLATION-SENSITIVE ASSETS
UHNW investors allocations to mining and energy projects are harder to execute, given that
the size of capital required to develop these assets is typically very large.
434
i
i
i
perrucci 2012/7/11 17:12 page 435 #457
19
Stefania A. Perrucci
New Sky Capital
Because general price levels affect asset returns and the economy as a whole, inflation is an important risk that every investor
should carefully analyse, and actively manage, in order to protect real wealth. In fact, the rationale for managing inflation risk
and for evaluating investments in real returnrisk space does not
need justification, as it is grounded in common sense. Instead, it
might seem surprising that virtually all commonly used investment metrics rely on nominal measures, which do not explicitly
capture the insidious impact of inflation on wealth. This cavalier attitude can be in part traced back to inflation-linked instruments (which are explicitly and formulaically indexed to inflation)
being a relatively recent innovation to fixed-income markets,1 combined with the intrinsic difficulty in assessing the correct dynamic
relationship between inflation sensitive assets (such as commodities, real estate, equity or infrastructure investments) and inflation
itself.
Besides being an important macroeconomic risk, inflation also
offers attractive opportunities to the active investor. Indeed, market inflation expectations displayed equity-like volatility during and
after the 20089 global crisis. This coincided with a substantial reduction in risk capacity on behalf of market makers, thus creating pricing
distortions and relative value opportunities due to supplydemand
shocks and hedging activities, in a market where liquidity can at
times be limited, despite the impressive growth since the tail end of
the crisis.
435
i
i
i
perrucci 2012/7/11 17:12 page 436 #458
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 437 #459
Insurance
companies IL
liabilities
Infrastructure IL
debt (UK,
Australia,
Canada)
80 bn
Utility companies
IL debt (Europe,
Australia, Brazil)
90 bn
Commercial real
estate IL debt
(Europe,
Australia)
40 bn
IL mortgage debt
(Iceland, Chile,
Mexico, Brazil,
Israel)
<30 bn
1,800 bn
1,400 bn
Asset managers,
hedge funds IL
assets
700 bn
Sovereign funds,
central banks IL
assets
300 bn
IL savings
accounts
100 bn
i
i
i
perrucci 2012/7/11 17:12 page 438 #460
INFLATION-SENSITIVE ASSETS
2.0
1.5
1.0
0.5
0
Dec 1971
Dec 1973
Dec 1975
Dec 1977
Dec 1979
Dec 1981
Dec 1983
Dec 1985
Dec 1987
Dec 1989
Dec 1991
Dec 1993
Dec 1995
Dec 1997
Dec 1999
Dec 2001
Dec 2003
Dec 2005
Dec 2007
Dec 2009
Dec 2011
2.5
i
i
i
perrucci 2012/7/11 17:12 page 439 #461
The sovereign inflation-linked debt market has increased dramatically to a size of almost US$2.5 trillion, with more and more
countries adopting a regular issuance schedule. This occurred over
a time when major economies experienced low, stable (or decreasing) inflation rates. In other words, it is not the risk of high inflation that has spurred interest in inflation-linked debt (and thus
promoted issuance) in the following two decades, but rather the
demand for a low-volatility investment asset, and effective portfolio
diversification.
However, with the increase in inflation volatility since the start of
the 20089 global financial crisis, new demand for inflation-linked
products has developed, with more investors taking an interest in
the space and its renewed opportunities. Incidentally, this demand
is likely to deepen the shortage of inflation supply (also shown in
Table 19.1), and thus provide good funding opportunities to natural
inflation payers, including non-sovereign issuers, particularly in the
infrastructure and utilities sectors.
Historically, the sourcing of inflation from non-sovereign issuers
has been limited, with the exception of the UK and, to some extent,
Australia (this is in line with pension fund hedging activity, which
is well developed in these two countries). However, as other markets evolve along similar paths, accounting/regulatory obstacles are
removed and assetliability management programmes encouraged,
we can expect non-sovereign issuers (whose financing needs we estimate to be about US$1.8 trillion globally) to progressively gain a
more prominent place among the global suppliers of inflation-linked
cashflows.
As for sovereign issuers, the context of large funding needs at the
time of writing points to a likely increase in inflation-linked bond
markets. However, governments will have to balance conventional
and real issuance, in order not to compromise the liquidity of their
nominal debt markets.
Inflation demand
Traditionally, pension funds have been the main driver of demand
for inflation products. This stems from the linkage of their liabilities to inflation, which can be either implicit (eg, taking the form of
annuities calculated from last salaries) or explicit (eg, in the form of
inflation-linked annuities, as in the UK). In some countries, such as
439
i
i
i
perrucci 2012/7/11 17:12 page 440 #462
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 441 #463
Market impact
Currency
BEI
Real yield
ASW discount
Option
US dollars
Euro
Sterling
Medium
High
High
Medium
High
High
Medium
High
High
Medium
Medium
High
i
i
i
perrucci 2012/7/11 17:12 page 442 #464
INFLATION-SENSITIVE ASSETS
2008. Hedging activity on the asset side provides support to longterm real rates, typically resulting in a relatively flat curve. Finally,
option overlay strategies generally provide support to out-of-themoney cap or floor inflation volatility, with large buyers at times
producing clear distortions in the volatility smile.
These technical flow dynamics are very powerful and cannot be
ignored, as they clearly exert a strong influence on market pricing
in sector.
Asset managers, hedge funds and sovereign funds have also
entered the inflation market, with important investment allocations
in the space. This source of demand is also on the rise, as alpha opportunities arise in the new climate of higher volatility, and the necessity
for managing beta inflation risk (de facto embedded in many portfolio) is increasingly recognised. Most large asset managers are longonly unleveraged investors, typically with a long-term horizon, who
are looking at locking-in attractive real rates. Some of the large asset
managers and sovereign funds are also benchmark sensitive. Retail
investors also participate in the space, and might be an important
price driver in certain situations, eg, when a specific inflation-linked
security is rebalanced out of a benchmark, so that it is typically sold
from large institutional portfolios.
As Table 19.1 makes clear, the potential demand for inflation products is large and at the time of writing cannot be met by the current
stock of inflation-linked bonds, despite their tremendous growth in
recent years. It is likely that, over time, this demand will be a driver of
additional supply, from sovereign and non-sovereign issuers alike,
as well as a main force behind further developments in the inflation
derivatives market.
The role of market makers
The role of market makers in mediating inflation flows is very
important, and understanding their structural positions and hedging activities is crucial in order to understand price behaviour in this
market.
Historically, inflation market makers, ie, investment banks, have
been structurally short BEIs via swaps (mainly sold to pension
funds in Europe) and short inflation volatility (mainly as a result
of coupon floors embedded in structured notes sold to the retail
market in Europe). These positions, and the strategies employed by
442
i
i
i
perrucci 2012/7/11 17:12 page 443 #465
i
i
i
perrucci 2012/7/11 17:12 page 444 #466
INFLATION-SENSITIVE ASSETS
US BEI (%)
DJIA (%)
Initial
Low
High
Range
2.60 = 100
11,671 = 100
17
4
12
10
29
14
i
i
i
perrucci 2012/7/11 17:12 page 445 #467
US 10Y BEI
EU 10Y BEI
0.5
Jul
2004
Jul
2005
Jul
2006
Jul
2007
Jul
2008
Jul
2009
Jul
2010
Jul
2011
its wake. Indeed, looking beyond the extreme gyrations experienced in 20089, and taking 2011 as an example, BEI swung up,
and then down, in synchronised moves, within a range that is
quite wide (29%), and more than double that of an equity index
such as the Dow Jones Industrial Average (14%) (Table 19.3 and
Figure 19.2).
Another consequence of the crisis is increased awareness of inflation risk on banks trading books, which, combined with new global
financial regulations and higher capital requirements, resulted in
limited balance-sheet and risk capacity on behalf of market makers.
At the same time, demand for inflation cashflows by large institutional investors has, if anything, increased, due to high volatility
in the market and also as a result of new regulations that focus on
inflation risk in a more explicit way. For example, Solvency II sets
capital requirements for insurance companies at the balance-sheet
level; thus, the effect of inflation on assets and claims makes capital
requirement inflation dependent. Because of these factors, the ability
to provide liquidity and intermediation of inflation risk and flows
structurally has decreased after the crisis, while demand has actually increased. This, in turn, has provided attractive opportunities
to the active manager.
445
i
i
i
perrucci 2012/7/11 17:12 page 446 #468
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 447 #469
i
i
i
perrucci 2012/7/11 17:12 page 448 #470
INFLATION-SENSITIVE ASSETS
25
UK
20
15
10
Inflation (%)
Japan
5
0
1961
1971
1981
1991
2001
5
2011
t IIN[0, 2 ]
i
i
i
perrucci 2012/7/11 17:12 page 449 #471
It is clear that if the matrix is a cointegrating matrix, for every nonsingular r r matrix M, then M will also be a cointegrating matrix.
449
i
i
i
perrucci 2012/7/11 17:12 page 450 #472
INFLATION-SENSITIVE ASSETS
Finding an economically sensible normalisation for the cointegrating space is part of the analysis (fixing the r2 terms in the matrix M,
ie, imposing just-identifying conditions). Often the number of conditions imposed on the model to ensure economic interpretability is
greater than r2 , resulting in a restricted model with over-identifying
conditions that can be tested and estimated accordingly.
It is customary to model inflation using a vector autoregressive
model of order p, ie, VAR(p). This is an n-dimensional model in
which the level of a stochastic vector process Xt depends on p lagged
values of the same vector
X t = At +
p
Bk Xtk + t ,
t IIN[0, ]
k =1
with constant covariance matrix . Note that we can have both a constant intercept and possibly linear or higher-order terms in At . The
above equation can be estimated by the ordinary least-squares (OLS)
method and then residuals checked for the multivariate normality assumption. The link with the cointegration approach becomes
clear if we consider the Granger Representation Theorem (Engle et
al 1987), which states that, for I (1) variables, a VAR(p) model can
be converted to a model combining both levels and differences (a
cointegrated model), ie, a vector error correction model VEC(q) of
order q = p 1
dXt = Xt1 +
q
Bk dXtk + t
k =1
Clearly, the first term must contain the stochastic equilibrium relationships among levels, while the second term is stationary by definition of I (1). Methods have been developed to test the existence and
number of cointegration relationships, among which is the Johansen
(1991) method. The latter infers the cointegration rank r by testing
the number of eigenvalues that are statistically different from 0 in
the error-correction matrix of the cointegrated VAR model, then
conducts model estimation under the rank constraints. Note that the
rank test is based on simulated non-standard asymptotic distributions that depend on the form of the VEC model, and the deterministic terms in particular. Once the rank r has been determined, the
Johansen procedure gives the maximum likelihood estimate of the
unrestricted cointegrated r 1 vector Xt (up to a normalisation
matrix M).
450
i
i
i
perrucci 2012/7/11 17:12 page 451 #473
a more structural effect on the economy and prices (commodity or currency shocks, changes in productivity or competitive
landscape, etc). By definition, exogenous shocks cannot be predicted, but they can be simulated, which we do extensively
through scenario analysis.
In our view, a monetarist approach to inflation is complementary rather than competitive to an expectation-augmented Phillips
curve, so we do look at the link between inflation and money growth,
although we realise that, empirically, the relationship is weak for
time horizons shorter than two years. In fact, we use a combination
of the money equation, the expectation-augmented Phillips curve
and our proprietary version of the Taylor rule to try to establish a
451
i
i
i
perrucci 2012/7/11 17:12 page 452 #474
INFLATION-SENSITIVE ASSETS
Money
equation
Expectation
augmented
Phillips
curve
Money
aggregates
Output
gap
Supply shocks
Inflation
expectations
Output gap
Taylor
rule
Inflation gap
general landscape that can then be explored in detail through specific scenario analysis (see Figure 19.4 for the key inputs to these
models).
While a bottom-up approach is used to gain insight on likely
short-term inflation behaviour (and it might be applied in practice
when estimating carry in short-maturity linkers), a top-down structural approach is typically used to establish long-term trends, and
in scenario/response analysis.
Monetary policy models
The Taylor Rule, introduced in 1993 as a descriptive model of monetary policy in the USA, links the US Federal Reserve funds rate to
the real economy and inflation
Fed funds rate = + GAP output gap + DEFL deflatorGDP
New Skys descriptive model of monetary policy, while belonging
to the family of Taylor Rules models, is in fact a proprietary model,
which provides an edge to our investment process. In fact, while the
Taylor rule (Figure 19.5) misses the tightening cycle of the mid-1990s,
and later the extent of easing of the post-tech-bubble recession, with
our proprietary definition of output gap, we are able to achieve a far
452
i
i
i
perrucci 2012/7/11 17:12 page 453 #475
10
8
6
4
2
Jan 2010
Mar 2011
Nov 2008
Jul 2006
Sep 2007
May 2005
Jan 2003
Mar 2004
Nov 2001
Jul 1999
Sep 2000
May 1998
Jan 1996
Mar 1997
Nov 1994
Sep 1993
Jul 1992
Mar 1990
May 1991
Jan 1989
superior fit than the traditional approach. Our framework also sheds
light on why the Feds have seen the need for quantitative-easing
policies, as the current slack in capacity would call for a negative
Fed funds rate.
Granted we do not have control on how the Feds set short-term
rates, we do have a pretty decent grasp of how the process has
worked in the past 20 years or so. Note that, having a good descriptive model of monetary policy is quite useful, but reading the minds
of policymakers does not in any way address the adequacy of the
monetary policy process itself.
Growth and real rates
As discussed in Chapter 7, when it comes to the inflation-linked
bond market, the primary focus of both domestic and international
(non-leveraged) investors is the level of real yields. Therefore, it is
natural to compare real rates, as a market-determined measure of
investment growth, with other benchmarks of economic growth,
such as the rate of change in real GDP.
At New Sky, we look at the link between real yields and real GDP
(the latter being a proprietary blend of realised and potential growth
rates), with the understanding that this is an equilibrium structural
relationship, which cannot be used to gain insight on short-term
453
i
i
i
perrucci 2012/7/11 17:12 page 454 #476
INFLATION-SENSITIVE ASSETS
Figure 19.6 Ten-year US TIPS real yields and real GDP growth
4.5
4.0
3.5
3.0
2.5
2.0
1.5
Ten-year real yield
0.5
1997 Q4
1998 Q2
1998 Q4
1999 Q2
1999 Q4
2000 Q2
2000 Q4
2001 Q2
2001 Q4
2002 Q2
2002 Q4
2003 Q2
2003 Q4
2004 Q2
2004 Q4
2005 Q2
2005 Q4
2006 Q2
2006 Q4
2007 Q2
2007 Q4
2008 Q2
2008 Q4
2009 Q2
2009 Q4
2010 Q2
2010 Q4
2011 Q2
1.0
i
i
i
perrucci 2012/7/11 17:12 page 455 #477
Cause
Scenario
description
trigger
Relative
Historical Real to target
example growth inflation
Policy
Cost push
Oil prices
shock
US
1970s
Below
Above
Loose
Demand
pull
Increase
in demand
China
today
Above
Above
Loose
Japan
1990s
Brazil
early
1990s
Below
Below
Ineffective
Below
Above
Ineffective
Expectations Deflation
spiral
Money
Increase in
supply
money
supply
i
i
i
perrucci 2012/7/11 17:12 page 456 #478
INFLATION-SENSITIVE ASSETS
Jul
2005
Jul
2006
Jul
2007
Jul
2008
Jul
2009
Jul
2010
Jul
2011
Because of the liquidity premium (and the deflation floor), zerocoupon real rates are not directly observable from TIPS. However,
they can be extracted from zero-coupon inflation swaps using a noarbitrage relationship, which is model independent. Suppose that, at
time t, the zero-coupon inflation swap with maturity T = t + N (that
is, N years from now), is quoted at a rate equal to KN,t . This means
that at time T the swap buyer will receive an amount equal to the
swap notional multiplied by the realised return on the price index,
456
i
i
i
perrucci 2012/7/11 17:12 page 457 #479
Realised inflation
Market
maker
Pension
fund
Swap BEI
KN,t
TSY
reverse
repo
IL TSY
repo
Long ZC
IL TSY
yt,TIPS
yt,N
Short ZC
TSY
Hedging strategy
Et
T
IT
exp
It
rsN ds
= Et
T
(1 + KN,t )N exp
rsN ds
The left-hand side is the price at time t of the zero-coupon real bond
maturing at T, while the right-hand side is a constant times the Tmaturity nominal zero-coupon bond price at time t
R
R
N
Pt,T
= exp[yt,T
(T t)] = (1 + KN,t )N Pt,T
R
Therefore, at each point in time t, the real term structure yt,T
can
be inferred from the term structure of inflation swaps and nominal
zero-coupon bonds.
The effect of the relative difference in liquidity and financing costs
between nominal and inflation-linked bonds can be illustrated by
considering how a zero-coupon inflation swap might be hedged
with a combination of a zero-coupon inflation-linked bond and a
zero-coupon nominal bond. In Figure 19.8, it can clearly be seen
that the difference in swap and TIPS BEI comes from the difference in repo/reverse repo costs between the inflation-linked and
the nominal treasury
IL
TSY
KN,t = BEISWAP = yt,N yt,TIPS
= BEITIPS +rIL rTSY
+r r
457
i
i
i
perrucci 2012/7/11 17:12 page 458 #480
INFLATION-SENSITIVE ASSETS
Short IL
asset swap
Realised inflation
Pension
fund
yt,N
Libor + ASWTSY
Swap BEI
KN,t
Realised inflation
Market
maker
yt,TIPS
Swap
ctpty
Libor + ASWIL
Swap
counterparty
i
i
i
perrucci 2012/7/11 17:12 page 459 #481
i
i
i
perrucci 2012/7/11 17:12 page 460 #482
INFLATION-SENSITIVE ASSETS
1Y
1994
Q4
10Y
1997
Q4
2000
Q4
2003
Q4
2006
Q4
2009
Q4
IRPt,
where
= ln
1+
MtR+
1
= exp
2
MtR
t+
t
Ts s
MtR+ It
,
MtR I t+
It
R
P
Pt, Et
It+
COVPt
ds
t+
t
Ts
dBPs
t+
t
rsR
ds
i
i
i
perrucci 2012/7/11 17:12 page 461 #483
0.2
0.4
0.6
0.8
1.0
Jan 2012
Jul 2011
Jan 2011
Jul 2010
Jan 2010
Jul 2008
Jan 2008
Jul 2007
Jan 2007
Jul 2006
Jan 2006
Jul 2005
Jul 2004
1.4
Jan 2005
1.2
This underlines the fact that the market might price inflation risk
quite differently depending on the overall environment and the balance in risk aversion between investors who prefer fixed real returns
and investors who prefer fixed nominal returns. Specifically, if the
market is focused on the risk of high inflation, then nominal securities will be penalised by a positive risk premium. On the contrary,
if the market is concerned about deflation, then the nominal bond
will benefit from being the better deflation hedge, and the inflation
risk premium will be negative. This observation also hints at the fact
that, although the price of risk is typically modelled as a Gaussian
variable in affine models, higher moments than variance (skewness,
ie, asymmetry in inflation risk, in particular) might be important in
order to model the inflation risk premium realistically (Garcia and
Werner 2010). In fact, although the variance in inflation expectations,
as measured by the SPF (which also publishes spreads in quartiles
of the future inflation distribution), has increased since the Great
Recession, it does not help to explain the dynamics of the inflation
risk premium or indeed the changes in sign. Unusual circumstances,
such as quantitative easing by the Feds and sovereign credit woes in
Europe, are likely to have contributed to the bipolar pricing of inflation risk from 2008 onwards, as well as to long periods of negative
461
i
i
i
perrucci 2012/7/11 17:12 page 462 #484
INFLATION-SENSITIVE ASSETS
real rates and an increased correlation between BEI and the equity
market.
Carry calculations
Carry can be an important value metric especially for short maturity
linkers. To calculate carry, we have to specify a time horizon, a financing rate and, in the case of inflation-linked bonds, non-seasonally
adjusted inflation rate projections over the relevant time window (so
that linkers carry inherits the index seasonality as well, which will
be discussed in the next section). A precise carry calculation should
take into consideration coupon income received and reinvested over
the period, the passage of time, yield roll-down, financing costs and
inflation accrual for the linker. However, given that these calculation
typically apply to short-term horizons and yet uncertainty about
future inflation rates supersedes concerns of mathematical precision, we shall use a simplified formula to guide intuition and write
the bond carry over a horizon of n = 1, 2, 3 months as
n
12
n
n
CPIk
CTIPS (n) = (yTIPS rTIPS ) +
12 k =1 12
CTSY (n) = (yTSY rTSY )
i
i
i
perrucci 2012/7/11 17:12 page 463 #485
Coupon Yield
BEI
2.000
1.83
0.125 0.35
2.18
0.20
0.30
2.28
0.21
2.5
0.46
4.0
0.10
2.28
0.14
0.16
0.27
0.56
0.41
0.41
0.02
0.29
0.00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
i
i
i
perrucci 2012/7/11 17:12 page 464 #486
INFLATION-SENSITIVE ASSETS
BEI
Duration
US Tsy (%)
0.30
US TIPS (%) 1.50
1.80
BEI
BE yield
0.15
0.60
0.23
Monthly (%)
accrual is not seasonally adjusted, projections need to take into consideration all the seasonal fluctuations, which is not always an easy
thing to do. Indeed, while inflation and real rate changes are often
(but not always) positively correlated and tend to mitigate the risk of
longer maturity linkers, short-term fluctuations in realised inflation
due to seasonal patterns (or non-core volatile components) are very
important to the value assessment of short-term linkers. In addition, seasonal fluctuations cannot be ignored, as their size can, in
certain months, overwhelm other fundamental considerations (see,
for example, the magnitude of the seasonality effect for March in
Figure 19.12). Clearly, non-seasonally adjusted (NSA) price levels
are the only directly observable quantities. The reason to identify
seasonality factors explicitly stems from the desire to forecast NSA
series (which will in general contain trend, cyclical, irregular and
seasonal components) by extracting the regular predictable patters
(ie, seasonality) first, and then focus on the rest.
Consider a two-year (remaining maturity) linker with BEI equal
to 1.8%, which corresponds to an inflation accrual on notional of
about 1.8%/12 = 0.15% per month. Even assuming the market has
got the annual inflation rate correct, in any given month inflation
will fluctuate because of seasonality, and potentially other factors
as well. In other words, monthly inflation rates might still compound to give an effective annual 1.8% rate but are likely to have
a wigglier pattern, mirroring the seasonality factors shown in Figure 19.12 (which sum to zero). Specifically, suppose that there are
reasons (seasonality being one of them) to believe that, next month,
NSA inflation will come in at a 0.60% rate (this is about equal to
the monthly BEI rate plus the seasonality adjustment for March in
Figure 19.12). This implies that the two-year duration linkers yield
can widen 45/2 = 22.5bp before it underperforms the corresponding
464
i
i
i
perrucci 2012/7/11 17:12 page 465 #487
11
16
21
Maturity
26
High strike
6
At-the-money
Low strike
i
i
i
perrucci 2012/7/11 17:12 page 466 #488
INFLATION-SENSITIVE ASSETS
Aug 2009
Sep 2009
Oct 2009
Nov 2009
Dec 2009
Jan 2010
Feb 2010
Mar 2010
Apr 2010
May 2010
Jun 2010
Jul 2010
Aug 2010
Sep 2010
Oct 2010
Nov 2010
Dec 2010
Jan 2011
Feb 2011
Mar 2011
Apr 2011
50
i
i
i
perrucci 2012/7/11 17:12 page 467 #489
Barter economy
Money economy
there are storage costs or a greater utility to later, rather than current,
consumption).
In the money economy, zero-coupon bonds denominated in basR
kets of goods, such as Pt,T
, are not tradeable instruments but their
US dollar value is (this is, in fact, our well known inflation-linked
bond, if one puts indexation lags and deflation floor aside). In other
words, to transition from the barter economy to the money economy, we need to multiply any basket-denominated quantity by the
price index It , which is denominated in US dollars per basket, thus
obtaining a dollar-denominated quantity.
Jarrow and Yildirim (2003) introduce three correlated Brownian
motions driving the stochastic dynamics of nominal and real instantaneous forward rates (so this model belongs to the HeathJarrow
Morton class), and the price level index It under the nominal data
probability P
R,N
T
Pt,T = exp
f R,N (t, u) du
t
0
R,N (s, T ) ds +
t
0
467
i
i
i
perrucci 2012/7/11 17:12 page 468 #490
INFLATION-SENSITIVE ASSETS
t
0
sI ds 2 tI2 +
t
0
sI dBIP
s
I
NI
dBN
t dBt = dt,
dBIt dBRt = RI dt
R,N (t, T ) = 0
= exp
t
0
rsN ds
the dollar value of real zero-coupon bonds It Pt,R and the dollar
value of the barter economy numeraire It BRt . Therefore, to avoid
arbitrage, there should be a probability measure Q under which
all our tradeable assets, discounted by the nominal numeraire, are
martingales
N
Pt,T
It Pt,R It BRt
,
, N martingales under Q
BN
BN
Bt
t
t
Using the link between forward rates and bond prices and the
dynamics equations for rates and the inflation index, by using Its
formula, we can derive the stochastic differential equations for the
three quantities above. From these, the change in measure that eliminates drifts, thus taking us to the risk-neutral probability Q, can be
inferred.
In the risk-neutral measure Q, nominal and real zero-coupon bond
prices as well as the price index have a lognormal distribution
dIt
IQ
= (rtN rtR ) dt + tI dBt
It
N
dPt,T
N
Pt,T
R
dPt,T
R
Pt,T
NQ
468
i
i
i
perrucci 2012/7/11 17:12 page 469 #491
where
R,N (t, T ) =
T
t
R,N (t, u) du
The model, by assuming lognormal bond prices (ie, normally distributed rates), has the drawback of allowing instantaneous and forward rates to go negative from time to time, but, on the plus side, the
lognormal assumption provides analytical solutions for both bond
prices and inflation caps and floors. For example, the price at time
t of the zero-coupon inflation cap, struck at K and with maturity T,
can be written as an expectation under the risk neutral measure
Q
where
IT = It exp
T
t
T
Ct = E t
max(IT K, 0) exp
(rSN rSR ) ds
1
2
T
t
rsN ds
sI2 ds +
T
t
sI dBIQ
s
i
i
i
perrucci 2012/7/11 17:12 page 470 #492
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 471 #493
2.71
2.6
2.5
2.4
2.3
2.2
2.1
2.29
Dec 2
Dec 16
Dec 30
Jan 13
Jan 27
Feb 10
Feb 24
Mar 10
Mar 24
Apr 7
Apr 21
May 5
Ma7 19
Jun 2
Jun 16
Jun 30
Jul 14
Jul 28
Aug 11
Aug 25
Sep 8
Sep 22
Oct 6
Oct 20
Nov 3
Nov 17
2.0
During the same period, the 10-year BEI rate held pretty steady,
and our estimate of the inflation risk premium stood solidly in positive territory, actually above the average of the conditional probability distribution. This and the overall macroeconomic picture suggested that a short BEI position had substantial upside in the case of
a regime shift (which we judged likely), and much less of a downside in case no such switch would occur. In reality, in the course of
the next 10 weeks, not only were BEI rates re-priced substantially
lower, but the inflation risk premium did in fact turn negative (Figure 19.11), as the market transitioned to pricing downside inflation
risk. The announcement of Operation Twist on September 21 also
contributed to a further step down in BEI rates, as nominal treasuries
rallied considerably after the news.
As it turns out, we entered the short BEI position on August 1 at
2.71%, and exited on October 6 at 2.29% (the actual trading pattern
was not linear, as technical effects also played a role).
Example: arbitrage trade
As explained earlier, swap BEI rates are typically higher than cash
BEI, mostly due to the difference in repo costs and ASW spreads
between nominal Treasuries and inflation-linked bonds.
471
i
i
i
perrucci 2012/7/11 17:12 page 472 #494
INFLATION-SENSITIVE ASSETS
Apr 2012
Feb 2012
Oct 2011
Dec 2011
Aug 2011
Apr 2011
Jun 2011
Feb 2011
Oct 2010
Dec 2010
Aug 2010
Apr 2010
Jun 2010
Feb 2010
Dec 2009
Oct 2009
Jun 2009
0.1
Aug 2009
i
i
i
perrucci 2012/7/11 17:12 page 473 #495
CONCLUSIONS
The inflation market provides a risk-transfer mechanism among
investors and hedgers. In this chapter, we explain how this market works, the different agents involved and what motivates them,
as well as the role of market makers in providing liquidity.
Inflation is not only an important macroeconomic factor but also
a risk that should be actively managed. Indeed, BEI rates displayed
equity-like volatility during the 20089 global crisis and have continued to do. This coincided with a substantial reduction in risk capacity on behalf of market makers, thus creating pricing distortions and
relative value opportunities.
In the first part of the chapter, we introduced some of the tools
essential to the analysis of the inflation market, including top-down
macroeconomic models and bottom-up pricing/market models. We
also described our investment philosophy and objectives, as well
as our holistic approach to the sector, which takes into account both
fundamental and technical considerations. Finally, we presented two
real-life examples of inflation strategies and the thought process/
analysis behind their evaluation.
APPENDIX: OIL PRICES SHOCK SCENARIO
This scenario might be relevant in a situation of renewed geopolitical
tension (we used this during the Arab Spring in 2011). A spike in oil
prices, if sustained, will translate in an increase in global headline
inflation. We shall focus on the impact in the US. In building this
scenario, we consider the direct effect on headline inflation (energy
being about 9% of CPI, although other items in the basket will also
be affected) and also on consumption (higher oil prices as an indirect
tax on consumers). We also assume a negative impact on real growth
and GDP.
We use a short-term model of inflation (Phillips curve and bottomup CPI-component analysis; see Figure 19.18), and output gap, after
which we let structural mean-reverting processes take hold. The path
of the short-term nominal rate is given by the Taylor rule (other
assumptions can be made). The evolution of inflation (swap) BEIs is
given by expectation plus a risk premium. Real rates are given by the
Fisher equation plus risk premium/covariance effect. As for monetary response, we expect the Feds to tighten in response, although
the rise in rates will be limited by the still negative output gap (see
473
i
i
i
perrucci 2012/7/11 17:12 page 474 #496
INFLATION-SENSITIVE ASSETS
Figure 19.18
Spike in oil
prices and
CPI; growth
slows down
Feds raise
rates; output
gap remains
negative
Inflation BEs increase;
inflation linkers rally;
nominal bonds sell-off;
pressure on currency
and equity
i
i
i
perrucci 2012/7/11 17:12 page 475 #497
i
i
i
perrucci 2012/7/11 17:12 page 476 #498
INFLATION-SENSITIVE ASSETS
fact that the focus on core inflation measures has worked well, in
the sense that core inflation has remained within range, irrespective
of commodity shocks, and with very limited monetary response to
those shocks (this is not true in the pre-Volcker era; see Evans and
Fisher (2011)).
In our opinion, however, the argument is dangerously backwardslooking, especially when we consider the difference in secular macro
trends at play, in particular the demographics of higher consumptions in emerging markets and the possibility that the latter might
turn from a deflationary to an inflationary global force. These secular changes may or may not play out (as usual, other effects such
as moderation of growth in emerging markets and/or technology
advances might counteract these forces). In any case, these effects
are never captured within the model, and if they materialise, they
will come as a surprise to many forecasters.
1
Refer to Chapters 7 and 20 for a brief history of inflation indexation in developed and emerging
countries respectively.
Keynes advocated the use of inflation-linked bonds in his testimony before the Colwyn
Committee on National Debt and Taxation in 1924.
Clearly, inflation-linked bonds also pay a semiannual coupon. Coupon cashflows can be
replicated by a portfolio of zero-coupon swaps and treasury strips, for each coupon date
(Fleckenstein et al 2010).
The five-year US TIPS Par ASW spread widened to 200 basis points over Libor. This was a
liquidity effect that did not affect nominal treasury ASW.
Examples of sudden shifts in inflation markets include those triggered by new regulations,
such as the indexation of tax-exempt saving accounts in France in 2003 (which created strong
demand for inflation-hedging products almost overnight) and the indexation of UK pension
annuities following the Pension Plan Act of 1995. Accounting regulations, especially those
pertaining to retirement benefits and their valuation (such as FRS17 in the UK), can also have
a profound effect on inflation players and markets.
See http://www.newskycapital.com/images/JUL11.pdf.
REFERENCES
Adrian, T., and H. Wu, 2010, The Term Structure of Inflation Expectations, Staff
Report 362, Federal Reserve Bank of New York.
Ang, A., G. Bekaert and M. Wei, 2008, The Term Structure of Real Rates and Expected
Inflation, The Journal of Finance 63(2), pp. 797849.
Chernov, M., and P. Mueller, Forthcoming, The Term Structure of Inflation Expectations,
Journal of Financial Economics.
Dickey, D., and W. Fuller, 1981, Likelihood Ratio Statistics for Autoregressive Time Series
with a Unit Root, Econometrica 49, pp. 105772.
476
i
i
i
perrucci 2012/7/11 17:12 page 477 #499
Engle, R. F., and C. W. J. Granger, 1987, Co-Integration End Error Correction: Representation, Estimation, and Testing, Econometrica 55, pp. 25176.
Evans, C., and J. D. M. Fisher, 2011, What Are the Implications of Rising Commodity
Prices for Inflation and Monetary Policy?, Chicago Fed Letter, May, p. 286.
Fleckenstein, M., F. A. Longstaff and H. Lustig, 2010, Why Does the Treasury Issue TIPS?
The TIPS-Treasury Bond Puzzle, Working Paper, UCLA.
Garcia, J., and T. Werner, 2010, Inflation Risks and Inflation Risk Premia, Working Paper
Series, No. 1162, European Central Bank.
Jarrow, R., and Y. Yildirim, 2003, Pricing Treasury Inflation Protected Securities and
Related Derivatives using an HJM model, Journal of Financial and Quantitative Analysis
38(2), pp. 337358.
Johansen, S., 1991, Estimation and Hypothesis Testing of Cointegration Vectors in
Gaussian Vector Autoregressive Models, Econometrica 59(6), pp. 155180.
Phillips, P. C. B., and P. Perron, 1988, Testing for Unit Roots in Time Series Regression,
Biometrika 75, pp. 33546.
Perrucci, S., 2009, Inflation: The Real Opportunity, URL: https://www.newskycapital
.com/.
Taylor, J. B., 1993, Discretion versus Policy Rules in Practice, Carnegie-Rochester Conference
Series on Public Policy, Volume 39, pp. 195214 (Elsevier).
Vasicek, O., 1977, An Equilibrium Characterization of the Term Structure, Journal of
Financial Economics 5, pp. 17788.
Wilson Committee, 1980, Report of the Committee to Review the Functioning of Financial
Institutions (Chairman Sir Harold Wilson), Command Paper 7937 (London: HMSO).
477
i
i
i
perrucci 2012/7/11 17:12 page 478 #500
i
i
i
perrucci 2012/7/11 17:12 page 479 #501
20
i
i
i
perrucci 2012/7/11 17:12 page 480 #502
INFLATION-SENSITIVE ASSETS
1969
1970
1972
1974
1976
1978
1980
1981
1983
1985
1987
1989
1991
1992
1994
1996
1998
2000
2002
2003
2005
2007
2009
2011
20
Australia
1%
Canada
3%
Japan
2%
Germany
2%
Emerging
markets
20%
US
35%
Italy 5%
France
9%
UK
22%
i
i
i
perrucci 2012/7/11 17:12 page 481 #503
i
i
i
perrucci 2012/7/11 17:12 page 482 #504
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 483 #505
From the 1960s to the 1990s, most Latin American countries experienced high inflation, and even hyperinflation, often caused by
the running of large structural fiscal deficits, which were financed
by printing money. In turn, excessive money growth resulted in
higher prices, and higher inflation expectations; the latter are an
important factor, as the de-anchoring of inflation expectations
explains the persistence of very high inflation in Latin American
countries for the greatest part of this period. According to Bernanke
(2005), the root of the problem was in the implementation of so-called
structuralist theories of development, which involve the protection of domestic companies from international competition. Domestic governments played an important role in this, by implementing
measures to prevent foreign competitors from entering local markets, and by heavily subsidising entire sectors of the economy. However, this came at the cost of chronic budget deficits, expansion in
monetary aggregates and mounting inflationary pressure.
Indeed, inflation indexation policies arose within the context of
high and persistent inflation, with domestic governments having
little choice but to issue inflation-linked debt as a way to secure
long-term funding in local currency. For example, after its creation
in 1953, the State Bank of Chile issued bonds to finance an ambitious
programme of development in the countrys agriculture and infrastructure sectors. Due to investors concerns about inflation and the
Chilean currency, these bonds were initially indexed to the US dollar. Then, in 1967, the authorities introduced a new unit of account,
the Unidad de Fomento (UF), tied to the ndice de Precios al
Consumidor (IPC), the Chilean Consumer Price Index. Such an
indexed unit of account has been the subject of several academic
papers (see, for example, Shiller 1998) and finds its roots in the
monetarist theory of Irving Fisher (1911). In the case of Chile, inflation indexation was soon applied not just to government bonds,
but also to a host of other financial transactions, including bank
deposits, residential rents, mortgage loans, house and commercial
property prices, alimony and child support payments and taxes.
Several Latin American countries followed a similar path. Ecuador
created the Unidad de Valor Constante in 1993; Mexico established
the Unidad de Inversion in 1995; Colombia created the Unidad
de Poder Adquisitivo Constante in 1972 and Uruguay adopted the
Unidad Reajustable in 1968.
483
i
i
i
perrucci 2012/7/11 17:12 page 484 #506
INFLATION-SENSITIVE ASSETS
From the 1990s onwards, faced with the negative impact of persistent high inflation, Latin American countries began to focus their
economic policies on the task of stabilising prices. Currency pegs
were introduced in several countries as a way to enforce fiscal discipline and anchor inflation expectations, a necessary condition for
long-term price stability. Mexico pegged its currency in 1987, as did
Argentina in 1991, while Brazil adopted a crawling band regime
in 1994. Although there were some early successes and inflation
rates decreased, currency-targeting policies could not be sustained
for a host of reasons. First, local governments in these emerging
countries lacked credibility in their fiscal and monetary policies. In
addition, currency pegs created barriers in the free movement of
capital across borders, thus affecting the stability of the domestic
economies. Furthermore, these pegged currencies were perceived
as over-valued and became the subject of intense market speculation. All of these factors forced governments to give up the currency
pegs (Mexico in 1994, Brazil in 1999 and Argentina in 2002) and
devaluate. Of course, this raised prices dramatically, but inflation
rates stayed below the peaks reached in previous decades. These
currency-devaluation episodes did not change the Latin American
countries focus on lowering and stabilising inflation. Following the
example of several developed (eg, New Zealand) and developing
economies (eg, Chile in 1990), most Latin American countries shifted
their monetary policy framework from exchange rate targeting to
inflation targeting. Peru did so in 1994, with Brazil, Mexico and
Colombia following in 1999, after which inflation rates did indeed
decrease and stabilise.
In reality, the successful reining in of inflation was not simply the
consequence of inflation targeting monetary policies, but the result
of a more complex combination of factors. As Ben Bernanke said
during a speech at the Stanford Institute:
I do not mean to claim, however, that Latin America conquered
inflation simply by choosing a particular framework for monetary
policy. Rather, my more fundamental point is that inflation has
declined in Latin America because new ideas and new political
realities have fostered the development of economic institutions
and policies that promote macroeconomic stability more generally. Recent changes in the policy environment have been especially
important in three areas: fiscal policy, banking regulation, and central bank independence. No monetary policy regime, including
484
i
i
i
perrucci 2012/7/11 17:12 page 485 #507
i
i
i
perrucci 2012/7/11 17:12 page 486 #508
INFLATION-SENSITIVE ASSETS
18
16
14
12
10
%
8
6
4
2
0
Jan 1980
Mar 1981
May 1982
Jul 1983
Sep 1984
Nov1985
Jan 1987
Mar 1988
May 1989
Jul 1990
Sep 1991
Nov1992
Jan 1994
Mar 1995
May 1996
Jul 1997
Sep 1998
Nov1999
Jan 2001
Mar 2002
May 2003
Jul 2004
Sep 2005
Nov 2006
Jan 2008
Mar 2009
May 2010
Jul 2011
i
i
i
perrucci 2012/7/11 17:12 page 487 #509
prices and the dependence of these countries on oil and food imports.
However, Asian economies are also evolving, with domestic demand
growing considerably and playing a stronger role as a determinant
of inflation. In fact, according to Osorio and Unsal (2011):
The relative roles of key inflation drivers appear to be changing
over time. The role of supply shocks in driving inflation appears
to have fallen slightly in recent years, while the role of output
gaps has increased. The impact of monetary shocks on inflation in
Asia has diminished, particularly in economies that have relatively
clear monetary objectives and flexible exchange rate regimes (such
as Indonesia, Korea, the Philippines, and Thailand).
Osorio and Unsal also mention that demand-driven inflation spillover effects from China to the rest of Asia are significant, and arise
from both higher imported goods and commodity prices. These
forces are likely to continue, thus contributing to both inflation and
its volatility in the foreseeable future.
However, despite the historically high and volatile inflation experienced in many Asian countries, the inflation-linked market is still
in its infancy, with the size of the outstanding inflation-linked debt
accounting for only a very small percentage of the overall total (Figure 20.4). South Korea and Thailand have recently issued inflationlinked bonds (Table 20.1), but again the size of issuance is underwhelming. This might be the consequence of the relatively small
size of the fixed income market in these countries, with local governments prioritising the development of the nominal rate market.
Another possible explanation is that pension fund schemes and the
insurance sectors are still at a fairly embryonic stage in many of these
countries; thus, the demand side has still to fully develop.
There are signs that Asian inflation-linked markets have the
potential to flourish, and room to expand, in the future. First, the
evolution and growth of pension schemes and the insurance sector
might take time but they are unavoidable, and they will create solid
structural demand for inflation-linked products. Indeed, when, in
2011, Thailand issued its first inflation-linked bond, private and public pension funds were the main investor targets. Secondly, although
practical obstacles to foreign investors entering the space (such as
withholding the tax that applies to South Korean inflation-linked
bonds) are still present, a lively debate has ensued in official circles as to how to eliminate such impediments and thus broaden the
487
i
i
i
First issuance
Inflation index
Israel
Brazil
1964
1964
1967
1967
1972
1989
1992
1997
2000
2007
2011
ISCPINM
BZPIIPCA
IGPIBREIGPM
COCPI
CLCPI
ARCPI
MXCPI
POCPILB
TUCPI
SACPI
KOCPI
THCPI
41
Colombia
Chile
Argentina
Mexico
Poland
Turkey
South Africa
South Korea
Thailand
Total
Sources: Deacon et al (2004), Bloomberg, Barclays Capital, New Sky Capital.
281
2
14
13
53
7
48
31
6
2
499
Country
INFLATION-SENSITIVE ASSETS
488
Table 20.1 Main issuers of inflation-linked bonds in emerging markets
i
i
i
perrucci 2012/7/11 17:12 page 489 #511
i
i
i
perrucci 2012/7/11 17:12 page 490 #512
INFLATION-SENSITIVE ASSETS
Argentina
3%
South Africa
6%
Poland
1%
Turkey
10%
Thailand
1%
Brazil
56%
Mexico
11%
South Korea
1%
Chile
3%
Colombia
0%
(sorted by first issuance date). Brazil has the largest stock of inflationlinked bonds (US$281 billion equivalent), followed by Mexico,
Turkey and Israel.
As seen in Table 20.1, Israel and Latin American countries pioneered inflation indexation and, as a result, these markets offer a
wider range of inflation-linked products, ie, bonds and derivatives
as well as several underlying indexes, and an array of domestic and
foreign investors, from pension funds to insurance companies and
asset managers, active in the space. Other countries are newer to
inflation indexation, and their markets are somewhat less sophisticated, with fewer products available, the latter typically being
sovereign inflation-linked bonds with a structure similar to US TIPS.
However, even these countries (one example being South Africa)
have been developing quickly in line with a parallel evolution of
their pension fund sectors sponsoring demand for these products.
At the same time, foreign demand has also risen considerably.
International institutional investors are actively allocating a portion of their assets to global inflation products, including emerging markets. Other emerging market investors are also considering
490
i
i
i
perrucci 2012/7/11 17:12 page 491 #513
491
i
i
i
perrucci 2012/7/11 17:12 page 492 #514
INFLATION-SENSITIVE ASSETS
Table 20.2 Brazil: federal public debt breakdown in 2011 and target
ranges
Target range
Type
2011
weights (%)
Lower
limit (%)
Upper
limit (%)
36.6
26.6
31.6
5.2
40
30
10
5
50
35
20
10
Fixed rate
inflation linked
Floating rate
Foreign currency
25%
15%
17%
41%
53%
24%
54%
20%
70%
24%
21%
63%
27%
52%
19%
2%
63%
4%
25%
11%
Financial Funds Pensions NonGovern- Insurers
institutions
residents ment
Other
Floating
Fixed-rate
33%
2%
Other
Inflation-linked
i
i
i
perrucci 2012/7/11 17:12 page 493 #515
IGP-M
Price index
Publisher
Period covered
Publication
IBREIGPM
Breakdown
Base
Seasonality
Non-seasonally
adjusted
currencies and foreign investment in fixed-income products is contingent to the payment of an upfront withholding tax (typically 1.5%
of notional). These factors are at the root of the developments of the
inflation swap market in Brazil, as we shall discuss later on.
As for the mechanics of sovereign Brazilian inflation-linked
bonds, several indexes and indexation methods have been used
since the start of the market in the 1960s. For example, in the early
1970s, indexation was based on a combination of both realised and
projected (by the government) inflation rates (Deacon et al 2004),
493
i
i
i
perrucci 2012/7/11 17:12 page 494 #516
INFLATION-SENSITIVE ASSETS
Figure 20.6 Brazilian annual inflation rates (IPC-A) and policy targets
YoY inflation
Inflation target
Jan 1999
Aug 1999
Mar 2000
Dec 2000
May 2001
Dec 2001
Jul 2002
Feb 2003
Sep 2003
Apr 2004
Nov 2004
Jun 2005
Jan 2006
Aug 2006
Mar 2007
Dec 2008
Jul 2009
Feb 2010
Sep 2010
Apr 2011
Nov 2011
20
18
16
14
12
% 10
8
6
4
2
0
Source: Bloomberg.
in an effort to minimise feedback effects from past to future inflation. Later, a fixed equilibrium rate replaced future projections. In
the mid 1970s, supply shocks corrections were introduced, which
effectively reduced government funding costs, but to the detriment
of investors.
As of spring 2012, two types of sovereign inflation-linked bonds
remain outstanding, the Notas de Tesouro Nacional-C (NTN-C) and
the NTN-B. The NTN-Cs, linked to the IGP-M index (Table 20.3),
were first issued in 1990 during the implementation of the Collor Plan I, aimed at stabilising the economy and inflation. Their
issuance was suspended in 1994 with the introduction of the Piano
Real, but resumed in 1999. The NTN-Bs, linked to the IPC-A index,
were introduced in 2002. At the time of writing they constitute most
of the inflation-linked debt outstanding (Table 20.4), as a result of
the Brazilian government focusing issuance on a single price index,
that is the IPC-A, in order to improve market depth and liquidity.
Indeed, the NTN-Cs share of inflation debt decreased from 75% in
2004 to 13% in 2011.
The choice of the IPC-A index is natural when we consider that
this is the same index adopted by the Brazilian Central Bank (BCB) in
conducting its inflation-targeting policy. This policy framework was
introduced in July 1999, six months after the BCB adopted a floating exchange rate system. Since then, the government has defined
inflation targets for the coming years (Figure 20.6) and given to BCB
494
i
i
i
perrucci 2012/7/11 17:12 page 495 #517
NTN-C
First issuance
2002
Outstanding (R, billions) 209
1990
25
Inflation index
Bloomberg index ticker
IPC-A
BZCLVLUE
IGPM
IBREIGPM
Auction
Up to two auctions
per month and swap
auction (buy short
term and sell longer
term)
Irregular
Coupon
Semiannual at 6%
Semiannual at 6% or
12%
Floor
Quotation
No
Nominal dirty
No
Nominal dirty
Business/252
15
2012,. . . ,2050
Business/253
3
Bloomberg ticker
BNTNB
i
i
i
Settlement date
Quoted price (nominal dirty price for a par value of 1,000)
(1) Base Bond Index (at bond issuance)
Latest IPCA: (April 2012)
Number of business days in the accrual period (from May 15 to June 15)
Number of business days accrued (from May 15 to May 21)
IPCA assumption (published by Andima)
(2) Bond index at settlement date: 3467.46 (1 + 0.46%)4/22 =
(3) Bond index ratio: (2)/(1)
Real dirty price = 2,870/(3)
Semi-annual coupon (1+6%)0.5 1
Number of days since the last coupon date
Number of days between two coupon dates
(4) Real accrued coupon: 2.956 64/125 =
Nominal accrued coupon: (4) (3)
Real clean price
Linker
INFLATION-SENSITIVE ASSETS
496
Table 20.5 From nominal dirty price to clean price
i
i
i
perrucci 2012/7/11 17:12 page 497 #519
i
i
i
perrucci 2012/7/11 17:12 page 498 #520
INFLATION-SENSITIVE ASSETS
BTUs
BCUs
401
354
and the Bonos del Banco Central de Chile (BCUs). The BTUs are
issued by the government, which uses the central bank as its agent.
In general, the government issuance follows a budget-stabiliser type
of strategy, which means an increase in net issuance when the economic activity is below its potential, and vice versa. As a result, the
issuance of BTUs decreased between 2000 and 2008, only to then
increase again afterwards. BTUs maturities range from 5 to 30 years.
The BCUs are issued by the central bank rather than the government. They play an important role in conducting monetary policy,
for example, in sterilised interventions, which aim at controlling the
foreign exchange rate.7 BCUs maturities range from 2 to 20 years.
Investors in BTUs and BCUs are typically domestic institutions, pension funds and insurance companies in particular. In general, taxation on interests and capital gains continues to be an obstacle for
most offshore investors (Barclays Capital 2012).
Similarly to sovereign bonds, the vast majority (92% of issues)
of corporate bonds are denominated in UF. In addition, derivatives instruments also trade with good liquidity. For short maturities
(within 18 months), forwards contracts on the UF index trade daily
via electronic platforms. In these contracts, one counterpart agrees
to pay the UF index in exchange for the variable short-term rate
Camara, which represents the average funding cost for domestic
financial institutions and is published daily by the Chilean Banking Association. For longer maturities (1Y, 5Y, 10Y and up to 30Y),
two types of real rate swaps are traded, with the real leg mirroring
the cashflows of inflation-linked bonds. One is the UF/Camara
swap, denominated in local currency. The other is the UF/Libor
swap, which settles in US dollars and has the quanto features we
previously hinted at when discussing Brazilian swaps.
Mexico
Mexico has the second largest stock (after Brazil) of inflation-linked
bonds in emerging markets countries. The first inflation-linked
498
i
i
i
perrucci 2012/7/11 17:12 page 499 #521
Coupon
Maturity
date
Issue
date
Outstanding
(Ps, billion)
3.25
5.5
3.5
4.5
5
3.5
4
2.5
4.5
4.5
4
Jun 2012
Dec 2012
Dec 2013
Dec 2014
Jun 2016
Dec 2017
Jun 2019
Dec 2020
Dec 2025
Nov 2035
Nov 2040
Jan 2009
Jan 2003
Jan 2004
Jan 2005
Jul 2006
Jan 2008
Jul 2009
Mar 2011
Jan 2006
Jan 2006
Mar 2010
7.2
9.0
10.2
15.1
10.8
13.3
10.2
11.8
8.6
24.3
20.3
Total
141.0
bonds were issued in 1996, two years after the Mexican pesos
crisis. Over time, with inflation stabilising and a nominal domestic debt market developing, the percentage of inflation-linked debt
has decreased from a peak of about 30% in the 1990s to about 20%
as of 2012. Given the importance of these instruments in meeting
demand from domestic institutions, and the governments focus on
supporting the evolution of a modern pension system, issuance of
sovereign inflation-linked bonds is here to stay.
In Mexico all inflation products are linked to the Unidad de
Inversion (UDI), which is an official index computed by the central
bank every two weeks. This index tracks the changes in the ndice
Nacional de Precios al Consumidor (INPC).8
Mexican sovereign inflation-linked bonds are called UDIBonos
(Table 20.7). Their structure is similar to the TIPS structure, where
a real coupon rate is applied to an UDI-inflated notional. The central bank, Banxico, acts as an agent of the Treasury and auctions
the bonds. To maintain a good level of market liquidity, the government runs a regular issuance schedule as well as auctions, where
long-dated bonds are exchanged for short-dated ones. Demand is
driven by domestic institutions, such as pension funds and insurance
companies, with long-term inflation-sensitive liabilities. Demand
from offshore investors, albeit on the rise at the time of writing,
499
i
i
i
perrucci 2012/7/11 17:12 page 500 #522
INFLATION-SENSITIVE ASSETS
i
i
i
perrucci 2012/7/11 17:12 page 501 #523
In the wake of these developments, the government funding strategy has gradually shifted on establishing a liquid nominal bond market. The share of inflation-linked bonds as a percentage of overall
government debt has been decreasing and at the time of writing it has
stabilised just above 40%. However, the demand for real rate assets
remain strong, as it is clear from the fact that about 70% of the corporate debt market is also indexed to inflation. According to Barclays
Capital (2012), the total stock of inflation-linked bonds in Israel was
NIS405 billion at the end of 2011, split between government inflationlinked bonds (NIS170 billion) and corporate inflation-linked bonds
(NIS235 billion).
Israeli Government inflation-linked bonds come in two flavours,
both linked to the Consumer Price Index (CPI). The Galil bonds have
both principal and interest indexed to the CPI and floored. Their
original maturities ranged between 2 years and 30 years, but the
longest outstanding issue has (at the time of writing) only 12 years
to maturity. Since 2006, a new type of inflation-linked bond has been
issued (also up to 30 years maturity), the main difference being
the absence of principal and coupon floors. This is partly due to
the historical large fluctuations of inflation rates in Israel, and the
governments reluctance to sell such deflation floors.
501
i
i
i
perrucci 2012/7/11 17:12 page 502 #524
INFLATION-SENSITIVE ASSETS
CONCLUSIONS
Several of the most advanced inflation-indexed markets are to be
found in emerging countries.
Indexation pioneers include Brazil and Israel, which issued their
first linkers in 1964, and Chile, which started issuance in 1967. To put
the timeline into perspective, this was about 20 years earlier than the
occurrence of similar developments in the UK, the first developed
country to start an inflation-linked programme in the modern era.
For many emerging markets, the issuance of inflation-linked
bonds was a piece of a much larger jigsaw. These countries had introduced a new inflation-linked accounting measure, which affected all
the aspects of economic activity from linking salaries to inflation to
mortgage and corporate debt. Indeed, these measures were aimed
at mitigating the effects of chronically high inflation, partly caused
by a structuralist economic approach, where the government subsidised industries, creating chronic deficit and high inflation in the
process. Indeed, inflation-linked bonds were one of the few viable
options for funding long-term local currency debt in these economic
circumstances.
In the 1990s, the focus of policymakers in emerging countries
turned to lowering and controlling inflation. After the failure of currency pegs in Mexico, Argentina and Brazil, inflation-targeting monetary policies were adopted in these countries and others. Their success had important consequences. The drop in inflation was coupled
with an anchoring of inflation expectations around the inflation target. Lower and more stable inflation expectations and less volatility
in inflation drove down market inflation break-evens significantly
across most countries. This initiated a de-indexation phase in several emerging market economies, and in countries such as Brazil,
Israel and Chile the focus gradually moved to the development of
the nominal yield curve. However, this did not mean the beginning
of the end of inflation markets. On the contrary, the rapid growth
and social development experienced by most developing countries
since the 2000s has seen the evolution of modern pension and insurance systems, which, together with important changes in the regulatory framework, has gradually created a large structural demand for
long-dated real return assets. Consequently, governments have been
reshaping their inflation-linked debt, through issuance or auctions,
eg, by extending duration in order to better match pension funds
502
i
i
i
perrucci 2012/7/11 17:12 page 503 #525
demand. This, together with increased market liquidity and simplified offer processes, has helped to lower long-term real yield and also
contributed to the development of a corporate inflation-linked bond
market in countries such as Israel, Brazil, South Africa and Chile,
whereas such markets remain limited in developed countries.
When it comes to inflation indexation, it is interesting to note how
the path followed by developed countries (the UK starting issuance
of inflation-linked debt in 1981, followed by the US in 1997 and
France in 1998) is quite different. These countries developed a nominal debt market first, and for them indexation was motivated by
the desire to capture the inflation risk premium, and thus achieve
better funding rates, rather than being dictated by chronic high inflation. Indeed, for most developed countries, the growth of sovereign
inflation-linked bonds has been exponential but occurred in a low
and stable inflation environment (with the exception of the UK). On
the demand side, the evolution of the market in developed countries
was driven in part by regulations that encouraged pension funds and
insurance companies to hedge their inflation-sensitive liabilities.
Despite the difference in historical and economic paths between
developed and emerging markets, at the time of writing we are witnessing a certain degree of convergence in global fixed income markets. After establishing a liquid nominal curve first, developed countries have, since the late 1990s, established or completed the real
curve as well. Conversely, several emerging countries, for which
indexation has traditionally been the only viable long-term funding option, have also established (or extended the maturity of) their
nominal programmes. In all cases, rising structural demand for
inflation-linked products (from pension funds and insurance companies) has been supporting indexation, often being a major cause
for issuance (eg, in South Africa and Thailand). We expect this to
continue to be a positive factor globally, and for Asian countries (eg,
India) in particular.
The gradual integration of emerging countries inflation markets
into the overall market is translating into more complex flows at the
global level, with new products (such as inflation focused mutual
funds and exchange-traded funds) being engineered in order to meet
new demand. Strong global demand for inflation-linked products
has translated into lower real rates, but more complex flows can also
plant the seeds for higher volatility, as the new sources of demand
503
i
i
i
perrucci 2012/7/11 17:12 page 504 #526
INFLATION-SENSITIVE ASSETS
can be more variable, especially compared with the traditional market activities of domestic pension funds and insurance companies.
Other macroeconomic factors, such as the rising volatility of commodity prices (an important factor, especially for emerging countries price baskets), add to the uncertainty, especially in the short end
of the inflation curve. Furthermore, the monetary policies adopted
in response to the 20089 recession, ie, exceptionally low rates and
quantitative easing, might come into question in the future and also
raise uncertainty about the future path of inflation globally.
All of these points make us believe that inflation markets will offer
opportunities as well as investment solutions to these very topical
issues. And, in this respect, the long history of inflation indexation
in emerging countries may indeed make them better prepared for
what is likely to be an uncertain future.
1
The UK Government started issuing inflation-linked bonds in 1981, the US Treasury started
in 1997 and France followed in 1998.
In 2010, State Street Global Advisors launched an ETF tracking the Deutsche Bank Global Governments Ex-US Inflation-Linked Bond Capped Index. This index measures the total return
of inflation-linked government bonds of both developed and emerging market countries outside the US. As of May 31, 2010, the index was composed of 17 government inflation-linked
benchmark indexes: Australia, Brazil, Canada, Chile, France, Germany, Greece, Israel, Italy,
Japan, Mexico, Poland, South Africa, South Korea, Sweden, Turkey and the UK.
The National Treasury of Brazil issues long-maturity (10Y, 30Y and 40Y) inflation-linked
bonds.
The exact calculation can be found in Federal Government Bonds: Methodology for Calculating Federal Government Bonds Offered in Primary Auctions at http://www.tesouro.gov.br/.
In most countries the inflation index is published with a lag. In addition, the index used for
settlement of inflation-linked bonds typically contains a lag and/or is an interpolated value
between lagged inflation index values.
Chile has had a floating exchange rate regime since 1999, but intervenes if the exchange rate
level deviates significantly from its equilibrium.
The INPC is released twice a month, on the 9th (accruing inflation between the 15th of the
month and the last day of the previous month) and on the 24th (accruing inflation between
the 1st and the 15th of the current month).
Foreign investors are not subject to withholding taxes or any other Mexican taxes (Barclays
Capital 2012).
REFERENCES
504
i
i
i
perrucci 2012/7/11 17:12 page 505 #527
Barnea, E., and N. Liviatan, 2008, The Chronic Inflation Process: A Model and Evidence
from Brazil and Israel, Journal of Economic Policy Reform 11(2), 151162.
Bnaben, B., and H. Cros, 2008, Global Inflation Derivatives Markets, in B. Bnaben and
S. Goldberg (eds), Inflation Risks and Products: The Complete Guide, Chapter 11 (London: Risk
Books).
Bernanke, B. S., 2005, Inflation in Latin America: A New Era?, Speech at the Stanford Institute for Economic Policy Research Economic Summit, Stanford, California,
February 11, URL: http://federalreserve.gov/boarddocs/speeches/2005/20050211/.
Chow, K., and R. Segreti, 2008, Inflation Products in Emerging Markets, in B. Bnaben
and S. Goldberg (eds), Inflation Risks and Products: The Complete Guide, Chapter 10 (London:
Risk Books).
Deacon, M., A. Derry and D. Mirfendereski, 2004, Inflation Indexed Securities (Chichester:
Wiley Finance).
Fisher, I., 1911, The Purchasing Power of Money (New York: Macmillan).
International Monetary Fund, 2004, Chile: Selected Issues, IMF Country Report 04/292.
Kaufman, G. G., T. H. Krueger and W. C. Hunter, 1999, The Asian Financial Crisis: Origins,
Implications and Solutions (Springer).
Osorio, C., and F. D. Unsal, 2011, Inflation Dynamics in Asia: Causes, Changes, and
Spillovers from China, IMF Working Paper WP11/257.
Poirson, H. K., 2007, Financial Market Implications of Indias Pension Reform, IMF
Working Paper WP/07/85.
Shiller, R., 1998, Indexed Units of Account: Theorey and Assessment of Historical
Experience, Cowles Foundation Discussion Paper 1171.
Tesouro Nacional, 2011, Optimal Federal Public Debt Composition: Definition of a LongTerm Benchmark, URL: http://www.tesouro.fazenda.gov.br/.
Tesouro Nacional, 2012, Federal Public Debt: Annual Borrowing Plan 2012, URL: http://
www.tesouro.fazenda.gov.br/.
505
i
i
i
perrucci 2012/7/11 17:12 page 506 #528
i
i
i
perrucci 2012/7/11 17:12 page 507 #529
Index
(page numbers in italic type relate to tables or figures)
A
ABP, 19
additive year-over-year inflation
swap, 126
affine term structure models,
21516
real and nominal, 24850
see also models
airports, 746
see also infrastructure assets
alternative calibration exploiting
currency analogy, 16970,
1734
American International Group
Commodity Index
(AIGCI), 18
arbitrage trade, 4712
Archer Daniels Midland
Company, 85
Art Market Research indexes, 427
assessment of liquidity in
inflation swaps and
options, 163
asset-class models to factor
models, 4026
asset prices:
and inflation, 27796
and money-neutrality
hypothesis, 2845
asset returns:
and inflation, 89
equilibrium-based
theories of, 3069
stock-price-based theories
of, 30913
and lagged inflation, 48
asset and sector rotation,
hedging through, 32546,
333, 334, 337, 341, 343
and broad asset classes,
impulse response, 33642
i
i
i
perrucci 2012/7/11 17:12 page 508 #530
INFLATION-SENSITIVE ASSETS
508
i
i
i
perrucci 2012/7/11 17:12 page 509 #531
INDEX
509
i
i
i
perrucci 2012/7/11 17:12 page 510 #532
INFLATION-SENSITIVE ASSETS
510
i
i
i
perrucci 2012/7/11 17:12 page 511 #533
INDEX
floors in inflation-linked
government bonds, 1567
food inflation, 3623, 362
forecasting inflation:
in developed versus
emerging markets, 3656
practical models for, 35167,
352, 354, 358, 359
in developed versus
emerging markets, 3656
and food, 3623, 362
fundamental, bottom-up,
35663
fundamental, top-down,
3516
and household energy,
3612
and petroleum, 35960
and shelter, 360
technical time-series,
3635
technical time-series models
for, 3635
forecasting and policy analysis
systems (FPAS), 18693
and forecasting horizons,
different models for,
1878, 187
major components in, 18891,
188
forward inflation curve:
granular, 1545
monthly, 1506
yearly, 150
FPAS, see forecasting and policy
analysis systems
Franco-Nevada, 86, 87
fundamental models for
forecasting inflation:
bottom-up models, 35663
top-down models, 3516
futures, and commodity price
indexes, 1416
G
global financial crisis, 345, 369,
460
511
i
i
i
perrucci 2012/7/11 17:12 page 512 #534
INFLATION-SENSITIVE ASSETS
Japan, 4, 302
Mexico, 302
UK, 4
US, 4, 302
annual, and policy targets, in
Brazil, 494
Art Market Research indexes
versus, 427
and asset prices, 27796
and money-neutrality
hypothesis, 2845
and asset returns, 89
equilibrium-based
theories of, 3069
stock-price-based theories
of, 30913
basic concepts of, 56
break-even, for, 10-year TIPS,
4078, 408
break-even, versus Dow
Jones Industrial Average
(2011) 444
in China, non-food, and
commodity prices, 29
and commodities, 2530, 26
and commodities shocks,
1314
commodity investing as
hedge against, 212; see
also inflation-sensitive
assets
and commodity investments,
379
compensation: expectation
and risk premium, 45962,
460
and CPI, see Consumer Price
Index
core, commodity prices
correlation with, 30
core, and money supply, 355
cumulative, 371
derivatives, 126, 41314
derivatives, with linear
payouts, 1258
different causes of, 68
duration, and real rate
duration, 399
512
i
i
i
perrucci 2012/7/11 17:12 page 513 #535
INDEX
513
i
i
i
perrucci 2012/7/11 17:12 page 514 #536
INFLATION-SENSITIVE ASSETS
514
i
i
i
perrucci 2012/7/11 17:12 page 515 #537
INDEX
515
i
i
i
perrucci 2012/7/11 17:12 page 516 #538
INFLATION-SENSITIVE ASSETS
516
i
i
i
perrucci 2012/7/11 17:12 page 517 #539
INDEX
year-on-year, complexity
adjustments in, 14550
and yearly forward
inflation curve, 150
zero-coupon, 1445
and options, assessment of
liquidity in, 163
and options, understanding
and trading, 13774, 165,
166
and currency analogy,
16770; see also currency
analogy
and development of
zero-coupon inflation
options, 1602
and hedging year-on-year
swap with two
zero-coupon swaps, 148
and historical and implied
volatility, comparison
between, 1646
and inflation-structured
notes, 1578, 157
zero-coupon, 1445
inflation targeting and emerging
markets, 260
inflation targeting and modern
monetary policymaking,
25762
inflationary periods, brief
history of, 3924
infrastructure assets:
airports, 746
defining, 6970
energy and water, 712
high leverage on,
implications of, 767
and inflation, 6977
impact of, 701
tollroads, 724, 72
insurance companies, impact of
inflation hedging by,
4412, 441
insurance portfolios:
protection of, from inflation,
36988, 374
asset allocation details, 376
517
i
i
i
perrucci 2012/7/11 17:12 page 518 #540
INFLATION-SENSITIVE ASSETS
and capital
adequacy/credit rating,
385
and decomposition of
variance of economic
value, 377
and different assets in
inflationary periods,
37981
and economic variance
decomposition, 3758
and efficient investment
frontiers for economic
value, 383
history and outlook,
36973
and impact on insurers
equity value, 3789, 378
and inflation-hedging
assets and efficient
frontier, 3814
and inflation-hedging
strategies, tailoring, 3846
and insurers equity value,
383
and investment income,
386
and liabilities, long-tail
versus short-tail, 385
liability composition
details, 377
and life insurance
companies, 3867
and risk tolerance, 386
and risk and return of
various asset classes, 382
simulation approach,
3745
interest, Fishers theory of, 299,
3034
interest rates:
Bank of Israel forecast
concerning, 183
forecasts of, drivers of change
in, 200
and inflation break-even
changes, effect of, 132
518
i
i
i
perrucci 2012/7/11 17:12 page 519 #541
INDEX
models:
affine term structure, 21516
affine term structure, real and
nominal, 24850
asset-class to factor, 4026
constant-volatility affine
AR(1), 216, 240
how used, 199
for inflation forecasting,
35167, 352, 354, 358, 359
in developed versus
emerging markets, 3656
and food, 3623, 362
fundamental, bottom-up,
35663
fundamental, top-down,
3516
and household energy,
3612
and petroleum, 35960
and shelter, 360
technical time-series,
3635
and interest-rate forecast,
drivers of change in, 200
and latent factors and
macroeconomic variables,
21920
multi-factor, of the short rate,
21415
multivariate, for near-term
forecasting, 2012
no-arbitrage, and expectation
hypothesis, 21213
option, and inflation markets,
4659
and pensions, 4026
pricing, and inflation
compensation in swap and
cash market, 4559, 456
projection, core, 1938
different types of, 1956
and parsimony, 193
and robustness to policy
errors, 1945
top-down approach to,
1934
519
i
i
i
perrucci 2012/7/11 17:12 page 520 #542
INFLATION-SENSITIVE ASSETS
520
i
i
i
perrucci 2012/7/11 17:12 page 521 #543
INDEX
521
i
i
i
perrucci 2012/7/11 17:12 page 522 #544
INFLATION-SENSITIVE ASSETS
522
i
i
i
perrucci 2012/7/11 17:12 page 523 #545
INDEX
ultra-high-net-worth investors,
42334
definition of, 424
inflation threat to, 4245
and legacy and time horizon,
425
and leverage, 4323
and liabilities, 4256
and real-asset life cycle,
42930, 430
and real assets, 4289
and risks and returns, 4302,
432
unemployment, and inflation, 7,
7, 3526 passim, 352
United Nations Food and
Agriculture World Food
Price Index, 30
US aggregate bonds, 335
US bond sector indexes, 3378
US Bureau of Labor Statistics, 5,
105, 331
US Federal Reserve, 25, 33, 185,
193, 195, 234, 256, 263, 314
Open Market Committee of,
191, 278
quantitative easing
programmes of, 372
unexpected steps taken by, 369
US Treasury Inflation Protected
Securities (TIPS), 10, 47,
5060 passim, 10513, 111,
113, 115, 2201, 220, 381,
454
and deflation floor, 232
indexation lag in, 2312
liquidity effects in, 22931
and real bonds and indexed
bonds, 2425
ten-year, break-even inflation
for, 4078, 408
zero-coupon yields, versus
real, 22932
see also bonds,
inflation-linked
523
i
i
i
perrucci 2012/7/11 17:12 page 524 #546
i
i