Download complete issue - IndexUniverse.com

We got your front.
Stop by the PowerShares booth
at the Inside ETFs Conference.

PowerShares ETFs give you 100% transparency.
All fund holdings are published daily, with intraday
pricing disseminated every 15 seconds. Because
when it comes to investing for your clients,
it helps to have all the facts.
The best situation for your clients is the agility to avoid a bad one.
// There are risks involved with investing
in exchange-traded funds (ETFs) including
possible loss of money. The funds are not
actively managed and are subject to risks
similar to stocks, including those related
to short selling and margin maintenance.
Ordinary brokerage commissions apply.
Shares are not FDIC insured, may lose
value and have no bank guarantee.
// Invesco PowerShares does not offer tax
advice. Investors should consult their own
tax advisors for information regarding
their own tax situations. While it is not
Invesco PowerShares’ intention, there is no
guarantee that the PowerShares ETFs will not
distribute capital gains to their shareholders.
// Shares are not individually redeemable
and owners of the shares may acquire those
shares from the Funds and tender those
shares for redemption to the funds in
Creation Unit aggregations only, typically
consisting of 50,000 shares.
// PowerShares ® is a registered trademark
of Invesco PowerShares Capital Management
LLC. ALPS Distributors, Inc. is the distributor
for QQQ. Invesco PowerShares Capital
Management LLC is not affiliated with ALPS
Distributors, Inc.
// An investor should consider the
Fund’s investment objective, risks,
charges and expenses carefully before
investing. To obtain a prospectus,
which contains this and other
information about the QQQ, a unit
investment trust, please contact
your broker, call 800.983.0903 or
visit www.invescopowershares.com.
Please read the prospectus carefully
before investing.
invescopowershares.com |
Follow us @PowerShares

fixing fixed income January / February 2012
Evaluating Alternative Bond Index Methodologies
Charles Thomas and Donald Bennyhoff
A Fundamentally Weighted Global Bond Benchmark
Shane Shepherd
Probability Of Default Metrics And Credit Indexes
Terry Benzschawel, Cheng-Yen Lee, Brent Hawker and David Craft
Perspectives On Fixed-Income Index Design
Brian Upbin
Plus Laipply and Woida on benchmark optimization,
Blitzer on blurring asset classes, Krein and Prestbo, Wiggins, Bell and more

www.journalofindexes.com
Vol. 15 No. 1
features
A Review Of Alternative Approaches
To Fixed-Income Indexing
By Charles Thomas and Donald Bennyhoff 10
Can market-cap weighting be improved upon?
A Fundamentally Weighted Broad-
Based Fixed-Income Index
By Shane Shepherd 20
Constructing a different kind of bond index.
Market-Implied Default Probabilities And Credit Indexes
By Terry Benzschawel, Cheng-Yen Lee,
Brent Hawker and David Craft 24
Building indexes based on probability of default metrics.
Optimizing Fixed-Income Index Funds
By Stephen Laipply and Christopher Woida 32
Trade-offs that can improve fund management.
Rethinking The Barclays Aggregate
By Rich Wiggins 38
The popular bond index comes with some drawbacks.
Bond Indexing Ripe For A Renaissance?
By David Krein and John Prestbo 40
Better data and investor interest create a flash point.
Bespoke Indexing
By David Blitzer 42
Next-generation indexes are blurring asset-class lines.
Perspectives On Fixed-Income Index Design
By Brian Upbin 44
Things to consider when choosing a benchmark.
Not Quite Rocket Science
By Heather Bell 66
An immodest proposal for fixing the Greek debt crisis.
news
S&P, Dow Jones Indexes Joining Forces . . . . . . . . . . . . . . 50
Indexing Developments............................. 52
Around The World of ETFs.......................... 54
Back To The Futures................................ 55
Know Your Options ................................ 55
From The Exchanges ............................... 55
data
Global Index Data ..................................56
Index Funds ........................................57
Morningstar U.S. Style Overview . . . . . . . . . . . . . . . . . . . . . 58
Dow Jones U.S. Industry Review . . . . . . . . . . . . . . . . . . . . . . 59
Exchange-Traded Funds Corner . . . . . . . . . . . . . . . . . . . . . 60
10
20
32
POSTMASTER: Send all address changes to Charter Financial Publishing Network, Inc., P.O. Box 7550, Shrewsbury, N.J. 07702. Reproduction, photocopying or
incorporation into any information-retrieval system for external or internal use is prohibited unless permission is obtained in writing beforehand from the
Journal of Indexes in each case for a specific article. The subscription fee entitles the subscriber to one copy only. Unauthorized copying is considered theft.
www.journalofindexes.com
January / February 2012
1

Contributors
Terry Benzschawel
Terry Benzschawel is managing director of Bond Portfolio Strategy at
Citi. He has a Ph.D. in experimental psychology and held multiple postdoctoral
fellowships prior to embarking on a career in finance in 1988 at
Chase Manhattan Bank. In 1992, Benzschawel joined Salomon Brothers,
which eventually became Citi. His focus throughout his career has been
on the risk and relative value of corporate and sovereign debt, with
recent emphasis on credit models.
David Blitzer
David Blitzer is managing director and chairman of the Standard & Poor’s
Index Committee. He has overall responsibility for security selection for
S&P’s indexes, and index analysis and management. Blitzer previously
served as chief economist for S&P and corporate economist at The McGraw-
Hill Companies, S&P’s parent corporation. A graduate of Cornell University,
he received his M.A. in economics from George Washington University and
his Ph.D. in economics from Columbia University.
Stephen Laipply
Stephen Laipply is a senior fixed-income investment strategist and a
member of BlackRock’s Model-Based Fixed Income Portfolio Management
Group. He joined the firm when his employer, Barclays Global Investors,
was acquired by BlackRock in 2009. At BGI, Laipply was a senior investment
strategist on the US Fixed Income Investment Solutions team. He
earned his B.S. in finance from Miami University, and his MBA in finance
from the University of Pennsylvania.
Shane Shepherd
Shane Shepherd is vice president and head of fixed-income research at
Research Affiliates. He conducts quantitative research used to strengthen
and expand the Research Affiliates Fundamental Index concept, and to support
the global tactical asset allocation model. Shepherd earned his Ph.D. in
finance from the University of California, Los Angeles. He holds a Bachelor
of Arts degree in political science and philosophy from Duke University.
Charles Thomas
Charles Thomas is an investment analyst for Vanguard’s Investment
Strategy Group. He conducts and supports research on a variety of topics,
including the global economy, fixed-income investing, currencies and
indexing. Prior to this role, Thomas participated in Vanguard’s leadership
development program, working in the retail and high-net-worth client
groups, and in ETF product management. He earned his B.A. in economics
from the University of Virginia.
Brian Upbin
Brian Upbin, CFA, CAIA, is a director in Barclays Capital’s Index Products group
and the global head of Benchmark Index Research and Product Development.
He joined Barclays Capital in September 2008 from Lehman Brothers, where
he was head of the U.S. Fixed Income Index Strategies team. Upbin received his
B.A. from the University of Pennsylvania, and his MBA from Yale University. He
is a member of the Fixed Income Analysts Society, Inc.
Rich Wiggins
Rich Wiggins is a senior consultant at Summit Strategies Group, based in
St. Louis. Summit is a leading institutional investment consultant ranked in
the top 20 by worldwide advisory assets. Prior to joining Summit, Wiggins
served as chief investment strategist at Citizens First Bancorp. He is a
past president of the CFA Society of Detroit and a periodic contributor to
Barron’s. Wiggins holds both the CFA and CAIA charters.
2 January / February 2012

Taxes can be
venomous to
wealth creation.
TAXES ON CAPITAL GAINS TAKE AWAY MONEY YOU COULD BE INVESTING.
WE STRIVE TO MANAGE OUR EXCHANGE-TRADED FUND (ETF) HOLDINGS IN A WAY
THAT MINIMIZES CAPITAL GAINS INCURRED BY THE FUND. SINCE OUR INCEPTION
IN 2003, THIS PROCESS HAS RESULTED IN ZERO CAPITAL GAINS DISTRIBUTIONS
ON 99% OF OUR ETFs. VISIT INVESCOPOWERSHARES.COM TODAY TO LEARN ABOUT
OUR TAX-EFFICIENT ETFs, INCLUDING POWERSHARES QQQ.
The best situation for your clients is the agility to avoid a bad one.
// There are risks involved with investing in
exchange-traded funds (ETFs) including possible
loss of money. The funds are not actively managed
and are subject to risks similar to stocks,
including those related to short selling and margin
maintenance. Ordinary brokerage commissions
apply. Shares are not FDIC insured, may lose
value and have no bank guarantee.
// Invesco PowerShares does not offer tax advice.
Investors should consult their own tax advisors for
information regarding their own tax situations. While
it is not Invesco PowerShares’ intention, there is no
guarantee that the PowerShares ETFs will not distribute
capital gains to their shareholders.
// Shares are not individually redeemable and
owners of the shares may acquire those shares from
the Funds and tender those shares for redemption
to the funds in Creation Unit aggregations only,
typically consisting of 50,000 shares.
// PowerShares ® is a registered trademark of
Invesco PowerShares Capital Management LLC.
ALPS Distributors, Inc. is the distributor for QQQ.
Invesco PowerShares Capital Management LLC is
not affiliated with ALPS Distributors, Inc.
// An investor should consider the
Fund’s investment objective, risks,
charges and expenses carefully before
investing. To obtain a prospectus, which
contains this and other information about
the QQQ, a unit investment trust, please
contact your broker, call 800.983.0903
or visit www.invescopowershares.com.
Please read the prospectus carefully
before investing.
invescopowershares.com |
Follow us @PowerShares

Jim Wiandt
Editor
jwiandt@indexuniverse.com
Heather Bell
Managing Editor
hbell@indexuniverse.com
Matt Hougan
Senior Editor
mhougan@indexuniverse.com
Lisa Barr
Copy Editor
Laura Zavetz
Creative Director
Jodie Battaglia
Art Director
Jennifer Van Sickle
Graphics Manager
Andres Fonseca
Online Art Director
Aimee Melli
Production Manager
Editorial Board
Rolf Agather: Russell Investments
David Blitzer: Standard & Poor’s
Lisa Dallmer: NYSE Euronext
Henry Fernandez: MSCI
Deborah Fuhr
Gary Gastineau: ETF Consultants
Joanne Hill: ProShare and ProFund Advisors LLC
John Jacobs: The Nasdaq Stock Market
Mark Makepeace: FTSE
Kathleen Moriarty: Katten Muchin Rosenman
Don Phillips: Morningstar
John Prestbo: Dow Jones Indexes
James Ross: State Street Global Advisors
Gus Sauter: The Vanguard Group
Steven Schoenfeld: Global Index Strategies
Cliff Weber: NYSE Euronext
Review Board
Jan Altmann, Sanjay Arya, Jay Baker, William
Bernstein, Herb Blank, Srikant Dash, Fred
Delva, Gary Eisenreich, Richard Evans,
Gus Fleites, Bill Fouse, Christian Gast,
Thomas Jardine, Paul Kaplan, Joe Keenan,
Steve Kim, David Krein, Ananth Madhavan,
Brian Mattes, Daniel McCabe, Kris
Monaco, Matthew Moran, Ranga Nathan,
Jim Novakoff, Rick Redding, Anthony
Scamardella, Larry Swedroe, Jason Toussaint,
Mike Traynor, Jeff Troutner, Peter Vann,
Wayne Wagner, Peter Wall, Brad Zigler
4 January / February 2012
Copyright © 2012 by IndexUniverse LLC
and Charter Financial Publishing Network
Inc. All rights reserved.

STOXX LIMITED
INNOVATIVE GLOBAL
INDICES
Innovation means, literally, “the introduction of something better” – a definition STOXX has always lived up to,
first and foremost with the EURO STOXX 50®, Europe’s leading blue-chip index. But STOXX has more to offer
than that: a truly global index family, ranging from blue-chip indices to broader benchmarks for virtually every
country and region. And because innovation is part of our DNA, STOXX offers advanced index concepts such
as Risk Control, Global ESG Leaders, and enhanced infrastructure indices – all applying a transparent, purely rulesbased
methodology. Besides being one of the financial world’s most renowned index brands, STOXX Ltd. also
markets and distributes the complete portfolio of DAX® and SMI Indices®, the leading brands from Deutsche
Börse AG and SIX Swiss Exchange AG. While innovation is our driver for product development, established
values such as trustworthiness and responsiveness are what matters most to us in client relationships. A successful
combination, as the figures clearly show: in 2010, approx. 185,000 structured products were issued on STOXX®,
DAX®, and SMI® indices, and our indices are the fourth-largest underlying for ETFs worldwide. More about the
index innovators on www.stoxx.com
STOXX is part of Deutsche Börse and SIX Group
STOXX®, EURO STOXX 50®, DAX®, and SMI® indices are protected through intellectual
property rights. The use of these indices for fi nancial products or for other purposes
requires a license from STOXX Ltd. (“STOXX”), Deutsche Börse AG (“DBAG”), and/or
SIX Swiss Exchange AG (“SIX”). STOXX, DBAG, and SIX do not make any warranties or
representations, express or implied, with respect to the timeliness, sequence, accuracy,
completeness, currentness, merchantability, quality, or fi tness for any particular purpose
of their indices and index data. STOXX, DBAG, and SIX are not providing investment
advice through the publication of the STOXX®, DAX®, and SMI® indices or in connection
therewith. In particular, the inclusion of a company in an index, its weighting, or the
exclusion of a company from an index, does not in any way refl ect an opinion of STOXX,
DBAG, or SIX on the merits of that company or qualifi es as an investment advice.
Contacts
Zurich, Headquarters: +41 58 399 5300 stoxx@stoxx.com

FREE subscRiption oFFER!
The Journal of Indexes is the premier source for financial index
research, news and data. Written by and for industry experts and
financial practioners, it is the book of record for the index industry.
To order your FREE subscription, complete and fax this form
to 732.450.8877 or subscribe online
at www.indexuniverse.com/subscriptions.
q Yes! Send me a free
subscription to Journal
of Indexes magazine
Foster Wright
Publisher
fwright@indexuniverse.com
646.867.4481 • Fax: 415.659.9005
Don Friedman
Director of Business Development
dfriedman@indexuniverse.com
415.659.9009 • Fax: 415.659.9005
Fernando Rivera
Reprint Sales Director
frivera@indexuniverse.com
646.588.6553 • Fax: 646.706.7051
Ivana Zivkovic
Subscriptions
izivkovic@indexuniverse.com
415.659.9029 • Fax: 415.659.9005
Signature
Name
Date
Dan Benveniste
Executive Vice President, Analytics & Data
dbenveniste@indexuniverse.com
415.659.9013
Title
Company
Address
City State ZIP
Phone
E-mail
The following best describes my primary business activity (check one):
(1) q Plan Sponsor (2) q Financial Advisor
(3) q Investment Management (4) q Mutual Fund Management
(5) q Pension Fund Consulting (6) q Pension Fund Management
(7) q Brokerage (8) q Academic
(9) q Ordinary Investor (10) q Other:__________________
Do you personally sell, recommend or manage investments, work in the index
industry or advise clients on investment and/or asset management?
q Yes q No
If you advise clients as to their investments, how do you charge them?
q Commission only q Fee only q Fee and commission
If you manage investments, what are your total assets under management?
(1) q Over $500 million (2) q $100 million - $500 million
(3) q $50 million - $99.9 million (4) q $25 million - $49.9 million
(5) q $10 million - $24.9 million (6) q Under $10 million
All questions must be answered to qualify for free subscription. Publisher reserves the right to reject unqualified applications.
Fax
Charter Financial
Publishing Network Inc.
499 Broad Street, Suite 120
Shrewsbury, NJ 07702
732.450.8866 • Fax 732.450.8877
Charlie Stroller, President/CEO/CFO
cstroller@fa-mag.com
IndexUniverse LLC
353 Sacramento Street, Suite 1520
San Francisco, CA 94111
1.877.6INDEX6 • Fax 415.659.9005
Jim Wiandt, President
jim_wiandt@journalofindexes.com
Charter Financial Publishing Network Inc. also
publishes: Financial Advisor magazine, Private
Wealth magazine, Nick Murray Interactive
and Exchange-Traded Funds Report.
For a free subscription to the Journal of Indexes,
IndexUniverse.com or Financial Advisor magazine,
or a paid subscription to ETFR, please visit
www.indexuniverse.com/subscriptions.
6
January / February 2012

Editor’s Note
A Fresh Angle
On Bond Indexing
Jim Wiandt
Editor
Clearly, bonds are no longer just for grandma. As equities market returns
have careened up and down in recent years with investors pouring ever
more cash into the perceived safety of fixed-income investments, innovations
designed to squeeze out a few additional basis points of income have come
on fast and furious in the index business. A wide array of alternatively weighted,
international and higher-yielding indexes have proliferated. In this issue, we
explore some of these innovations as well as calls for sanity.
We kick things off with a well-argued piece from Charles Thomas and Donald
Bennyhoff of Vanguard, who make the case for the traditional cap-weighted, broad
market approach to bond index investment over alternatively weighted approaches.
Shane Shepherd of Research Affiliates begs to differ, and explains how Research
Affiliates has constructed its own fundamentally weighted global bond index.
Terry Benzschawel and the team at Citigroup offer their own take on bond indexes,
describing their probability-of-default metric and how it can be used in fixedincome
construction, while Stephen Laipply and Christopher Woida of BlackRock
discuss some of the trade-offs portfolio managers must make when managing
index-based investment products.
Next up, Summit Strategies Group’s Rich Wiggins thinks you need to take another
look at the Barclays Capital U.S. Aggregate Index, and David Krein and John
Prestbo of Dow Jones Indexes show how the growing demand for fixed-income
investments has created a unique opportunity for index providers. S&P’s David
Blitzer offers his own perspective on new index strategies and how they’re blurring
the lines between asset classes, while Brian Upbin from Barclays Capital discusses
some fixed-income benchmark alternatives.
Heather Bell closes out the issue with some suggestions for solving that big bond
dilemma, the Greek debt crisis. Hint: Yogurt is involved.
Here’s to 2012 yielding you health, happiness and some decent returns.
Jim Wiandt
Editor
8
January / February 2012

There’s nothing passive
about how you invest.
You’re continuously fielding opportunities, adjusting allocations,
managing risk and most importantly, realizing your client’s
financial objectives. So how do you bring it all together?
Access ETFs or mutual funds that track S&P Indices.
Act on your ideas, knowing that S&P Indices’ transparent
methodology supports your choices.
UÊÊCore solutions. Build core strength with widely traded
investments that are linked to a time-tested lineup of
domestic and global equity indices.
UÊÊTactical solutions. Pinpoint tactical opportunities across
broad sectors, narrow industries and regional exposures
with precision, speed and agility.
UÊÊIncome solutions. Shape diversified income strategies that
capture investment-grade municipal bonds, preferred stocks
or high yield equities.
UÊÊReal asset solutions. Diversify with global commodity indices
that give easy access to real assets through commodity futures
or common stocks.
S&P Indices is your source for a full spectrum of solutions that
get ideas moving. www.spindices.com/financialadvisors
This information does not constitute an offer of services in jurisdictions where Standard & Poor’s does not have necessary licenses. Standard & Poor’s receives compensation
in connection with licensing its indices to third parties. It is not possible to invest directly in an index. Exposure to an asset class is available through investable
instruments based on an index. There is no assurance that investment products based on an index will accurately track index performance or provide positive investment
returns. Standard & Poor’s does not sponsor, endorse, sell, promote or manage any investment fund or other vehicle that is offered by third parties and that seeks to provide
an investment return based on the returns of any of our indices. For more information on any S&P Index and any further disclosures please go to www.standardandpoors.
com. Copyright © 2011 Standard & Poor’s Financial Services LLC, a subsidiary of The McGraw-Hill Companies, Inc. All rights reserved. STANDARD & POOR’S and S&P are
registered trademarks of Standard & Poor’s Financial Services LLC.

A Review Of Alternative
Approaches To
Fixed-Income Indexing
Can market-cap weighting be improved upon?
By Charles Thomas and Donald Bennyhoff
10
January / February 2012

The recent global financial crisis has led to heightened
sensitivity to the indebtedness of issuers across the
global fixed-income marketplace. The implications
of the events in Europe for both sovereign and corporate
issuers have led some investors to question the merits of
indexes constructed based on the traditional method of
market-capitalization weighting, according to which, larger,
more indebted issuers inherently receive a higher weighting
in the index. This paper evaluates several alternative indexweighting
methodologies commonly proposed in the global
sovereign and U.S. corporate marketplaces.
Beta, The Market And Indexing
Traditionally the term “beta” has been used to describe
the risk/return attributes of a particular asset class.
Accordingly, beta in the typical sense is synonymous with
the performance characteristics of “the market,” meaning
the total invested capital in a group of securities, such
as the stocks or bonds in aggregate. An index is a group
of securities chosen to represent the characteristics of a
particular market. Indexed investing (or indexing) is an
investment strategy designed to closely mimic the risk/
return attributes, or beta, of the benchmark index being
tracked. Since an index does not reflect all of the costs
and potential implementation hurdles that would make
it investable, the index is only a theoretical performance
benchmark. Indexing—via a mutual fund or exchangetraded
fund (ETF), for example—reflects these implementation
costs and, therefore, should provide investors with
a reasonable proxy for achievable or investable beta.
Once the market being measured is clearly defined, we
believe that a well-designed index will accurately reflect
the risk/return attributes of the total capital invested
by participants in that market, which can only be fully
accomplished through a market-cap-weighting process. 1
This is an important consideration, since it is these market
participants who set prices, incorporating all risks they
perceive relevant. By reflecting the value of all assets that
investors have chosen to invest in a particular market,
an index provides a benchmark for that market’s beta, as
well as a benchmark for how the average dollar (or yen,
euro, etc.) performed in that market.
Investment performance can be deconstructed into
three parts: the portions of return attributable to the
market (beta); to market-timing; and to security selection
[Brinson, Hood and Beebower, 1986]. The latter
two portions are specific to active management, while
the first can be measured through benchmark indexes.
As a result, we believe that the best index is not necessarily
one that provides the highest return over a given
period, but the one that most accurately measures
the risk/return characteristics of the collective capital
invested within the market being tracked.
Advantage Of Cap-Weighting
A cap-weighted index reflects the relative value of debt
securities as determined by market participants, without
changing the market’s relative composition. This makes
cap-weighting distinct from other weighting methods in
several important ways.
The key variable determining a security’s weight
in a cap-weighted index is its price. In public capital
markets, the market price of a security reflects every
market participant’s information, beliefs and expectations
regarding the value of that security. Price thus
represents a powerful mechanism collectively used by
market participants—often with very different opinions
and valuation processes—to establish and change
views on future performance.
In the global fixed-income marketplace, investors buy
and sell bonds based on a variety of factors, the most
important of which is the investor’s perception of the
issuer’s willingness and ability to honor the terms of the
obligation. As a result, default risk is a principal consideration
in fixed-income valuation and is reflected in
the prices of an issuer’s bonds. Continuous buying and
selling ensures that the price of any given bond reflects
the consensus estimate of its intrinsic value, accounting
for the expected risk and return from every investor’s
valuation process. In this sense, when a benchmark
is weighted by market cap, it is a reflection of all market
participants’ views regarding the relative value of
all fixed-income securities within that market. So we
believe it is inappropriate to view the benchmark as
being weighted by a single factor only—such as price—
or even by a limited number of factors. 2
Assumptions Of Alternatively Weighted Indexes
Recently, alternative index-weighting methods for various
sectors across the global fixed-income market have
been devised, perhaps to capitalize on the sentiment
associated with events of the recent global financial crisis
and ongoing fiscal issues in many developed markets.
Common alternative-weighting criteria include gross
domestic product (GDP), population and other macroeconomic
measures for sovereign debt; and assets, revenue
and other financial statement metrics for corporate issuers.
3 However, weighting based on one or several factors,
such as those used to form alternative weights, assumes
that the market-consensus valuation is wrong, and that
these other factors are better predictors of the fair value of a
security (see Figure 1). This seems questionable, given that
a rules-based index would have to be based on factors that
were publicly available, and as such should be reflected in
market prices. In any case, the index provider’s decision
to deviate from market weights inherently deviates from
that market’s performance characteristics (beta) and introduces
an aspect of active risk (security selection). 4
A key question, then, is what risk exposure do these strategies
entail to produce a given return? Although security
selection has the potential to enhance returns, systematic
tilts toward persistent risk factors within a given market (for
example, overweighting speculative-grade corporate bonds
in the U.S. bond market) can produce predictable and replicable
performance characteristics and may be best considered
a targeted beta strategy. The following sections discuss
www.journalofindexes.com January / February 2012 11

Figure 1 Figure 2
A Market-Cap-Weighted Index Is An 'All-Factor' Index
Alternative Indexing Underweights Japan And
Overweights Emerging Market Countries
Should an investor
focus on
some factors?
GDP
Landmass
Population
Energy Consumption
Dividends
Assets
Cash flow
Book value
Sales
Or do other factors
matter as well?
Political risk (willingness)
Inflation
Exchange-rate policy
Externally held debt
Composition of economy
Capital flows
Liquidity
Investability
Workforce productivity
Demographics
Profits
Competitive landscape
Management effectiveness
Corporate-governance controls
New products/lines/business
Regulatory environment
Off-balance-sheet items
Supply chains
Industry outlook
Market
capitalization
captures all
potential
factors that
all investors
collectively
use to
determine a
bond’s price
Distribution, by region, for weighting methods in global
sovereign debt market: May 31, 2011
Percentage Of Global Bond Market
35%
30%
25%
25%
22%
20%
15%
10%
5%
0%
15%
28%
9%
4%
11%
9%
4%
U.S. Japan Peripheral
Europe
19%
17%
Other
G-7
10% 10%
15%
11%
Other
Developed
Markets
6%
11%
BRIC
32%
5%
15%
24%
Other
Emerging
Markets
Any factor used by market
participants
Source: Vanguard, based on data from Barclays Capital and Arnott [2010]
Notes: “Cap-weighted” reflects the combined weights of the Barclays Capital Global
Treasury Index and Emerging Markets Government Universal Index. “GDP-weighted”
reflects the GDP-weighted versions of the same two Barclays Capital indexes. The
weights are adjusted to account for the overlap of countries in the two indexes.
“RAFI-weighted” reflects the methodology within Arnott [2010].
several alternative index strategies in the global sovereign
and U.S. corporate debt markets. We compare them with
the cap-weighted market, and we demonstrate that these
indexes frequently involve a meaningful reweighting of
common bond market risk factors.
Alternative Indexes In
The Global Sovereign Market
Common nonmarket-cap-weighted index methodologies
within the global sovereign marketplace have focused
on moderating exposure to more indebted countries by
using country-level metrics, including GDP, landmass
and population. This reflects the fact that sovereign-debt
issuers have the ability to levy taxes on the economy that
they govern, and are typically evaluated using broad,
macroeconomic measures. Some examples include GDPweighted
versions of the Barclays Capital Global Treasury
Index (mostly developed markets) and the Barclays
Capital Emerging Markets Government Universal Index. 5
Various industry publications have also featured discussions
of a RAFI (Research Affiliates Fundamental Index)
weighting system that combines several macroeconomic
factors, including GDP, landmass, population and energy
consumption. 6 Note that these alternatively weighted
indexes use widely available public data that should—to
the extent the data are relevant in determining intrinsic
value—be reflected in prices, and therefore market cap.
We evaluate these two alternative weighting methods
(GDP and RAFI) in the following section.
Although these two weighting methodologies (GDP
and RAFI) produce different weights, the implications
are the same: Both may produce large, persistent country
biases relative to the cap-weighted sovereign market
(see Figure 2). The country weights associated with the
alternatively weighted indexes are the result of a shift
away from countries that have high debt loads relative to
their alternative weighting factors and toward countries
with low debt loads relative to their alternative weighting
factors. For both weighting methods, the two most notable
differences versus the global-cap-weighted market
involve an underweighting to Japan and an overweighting
to the sovereigns of emerging market governments.
These country biases reflect the general appeal of
alternative weighting methodologies: They reallocate
weight away from governments with the highest relative
debt levels, as shown in Figure 3. The implicit assumption
here is that the bonds of a government with relatively
Figure 3
Government Debt Levels Across The World
Government debt as percentage of GDP: 2010
Percentage Of Region GDP (%)
250
200
150
100
50
0
100
66
Developed
Markets
94
68
220
11%
117 104
89 82
U.S. Japan Peripheral
Europe
Other
G-7
Emerging
Markets
BRIC
Sources: Vanguard and International Monetary Fund, September 2011,
World Economic Outlook
Notes: Net government debt is not available for all emerging market governments,
and so is not shown. U.S., Japan, Peripheral Europe, and Other G-7 country data are
subsets of “Developed Markets.” BRIC is a subset of “Emerging Markets.”
61
39 42
12
January / February 2012

more debt outstanding are poor investment prospects.
Indeed, governments of the peripheral economies of the
euro area, including Greece, Ireland, Italy, Portugal and
Spain, have been experiencing pressure from financial
markets precisely because of seemingly unmanageable
debt loads. Recent attention on the projected deficit
spending in the United States and other large developed
nations has led to a further focus on this issue.
At first glance, abandoning market weights may seem
to be an investment solution to avoid trouble spots in the
global sovereign market. However, investors should be
cautious about applying the experience of the peripheral
European countries more generally: A relatively larger
debt load does not always indicate that a given government’s
debt is a bad investment.
Does ‘Indebtedness’ Matter For Yields And Returns?
Focusing on one factor (“indebtedness”) may miss
other, more important things in determining sovereign
risk. As shown in Figure 4, since 1980, in a group of the
largest developed markets, government indebtedness (as
measured by debt to GDP) and yields have had a weak
relationship. Across the time period and across countries,
we find a wide range of both nominal and real yields associated
with any given level of government debt. In fact,
Japan’s presence in the nominal sample shows an opposite
relationship from what many might expect. Higher
debt levels were actually associated with lower yields,
because of Japan’s low inflation profile. After accounting
for inflation differences across these countries, as shown
in the real yields, government indebtedness had almost
no power in explaining yield levels. 7
Extending this view to returns, Figure 5 shows the relationship
of government debt levels and future real returns
Figure 4
Debt Levels And Yields Have A Weak Relationship
Bond yields versus government debt levels in developed markets: 1980−2010
Long Bond Yield (%)
20
15
10
5
0
-5
0
50
100
R 2 = 0.0378
R 2 = 0.1777
Sources: IMF and Thomson Reuters
Notes: Includes the G-7 countries plus Australia and Switzerland, annually from 1980
through 2010. The “nominal yield” series represents the average yield level of that
country’s ten-year maturity government bond within each year. The “real yield” series
adjusts that yield to real terms, using end-of-period consumer price inflation from the
IMF. Government debt is shown gross of financial assets, as defined by the IMF. The
results are similar using net debt.
150
Japan
Government Debt To GDP (%)
200
250
Figure 5
Debt Levels Do Not Predict Future Returns
Forward real returns versus government debt levels
in developed markets: 1985−2010
Annualized Real Returns
30%
25%
20%
15%
10%
5%
0%
-5%
-10%
-15%
0
25
50
75
100
R 2 = 0.0261
R 2 = 0.0199
Government Debt To GDP (%)
125 150 175 200 225
Source: Vanguard, based on data from Citigroup and IMF
Notes: Data include the G-7 countries plus Australia and Switzerland, from 1985
through 2010. Government debt is shown gross of financial assets, as defined by the
IMF. The results are similar using net debt. Returns are calculated based on the
Citigroup World Government Bond Index for each country, in local terms, adjusted to
real terms using end-of -period consumer price inflation from the IMF. Government
debt levels in a given year are compared with both the forward return in the first
calendar year following and over the five calendar years following. “1-year-ahead”
data cover periods as follows: debt-to-GDP (1985−2009); real returns (1986−2010).
“5-year –ahead” data cover periods as follows: debt-to-GDP (1985−2005); real returns
(1990−2010).
across the same group of developed markets, using market-weighted
sovereign-debt indexes to measure returns.
As in Figure 4, there is a weak relationship between this
debt and yields. This is because, when compared to other
factors (inflation, central bank activity, economic expectations,
to name a few), a government’s debt outstanding
may be relatively unimportant in determining appropriate
yields. 8 Many specific factors make each country and
government unique; thus, different governments are able
to sustain varying overall debt levels.
The experience of some emerging market countries provides
useful case studies for this point. In the late 1990s and
early 2000s, many emerging markets experienced financing
difficulties, with some eventually defaulting on their obligations,
despite carrying lower debt loads than many developed
markets. For example, Russia defaulted on its sovereign
debt in 1998, with externally held government debt
totaling about 30 percent of GDP in 1997, while Argentina
defaulted on its sovereign debt in 2002, with government
debt totaling about 55 percent of GDP in 2001. 9
In both instances, a significant contributing factor
undermining the ability of these issuers to repay their
debt was the collapse of a fixed-exchange-rate regime that
resulted in a depreciating national currency and appreciating
dollar-denominated government debt. The challenges
associated with fixed exchange rates and their role
in precipitating sovereign defaults have been well documented,
and the experience of Russia and Argentina is
not unique. 10 Governments with fixed exchange rates and
debt denominated in foreign currency (typically, emerging
markets) face different risks from those with floating
exchange rates and debt denominated in their national
currency (typically, developed markets). Exchange rate
www.journalofindexes.com January / February 2012 13

Figure 6
Figure 8
Emerging-Market Bonds Have Equitylike
Risk Characteristics
Alternative Index Weights Result In Lower Duration
Within Developed-Market Sovereigns
Correlation of sovereign debt to traditional asset classes: 1994−2011
GDP-Weighted Vs. Market
Correlation To Traditional Asset Classes
0.8
0.6
0.4
0.2
0.0
-0.2
0.25
0.15
–0.06
U.S.
Treasurys
0.11
0.44
0.24
U.S. Corporate
Bonds
–0.09
0.50
0.45
0.050.52
U.S. High-
Yield Bonds
Developed-
Market
Equities
0.73
0.65 0.66
–0.04
Emerging-
Market
Equities
Sources: Barclays Capital, JPMorgan and MSCI
Notes: Displays the monthly correlations of bond indexes relative to a group of
traditional asset classes over the period 1994−2011. “Japanese bonds” are defined as
the Barclays Capital Global Treasury Japan Index; “Emerging-Market Bonds
(USD-Denominated )” are defined as the JPMorgan Emerging Markets Bonds Plus
Index; and “Emerging-Market Bonds (Local-Denominated)” are defined as the
JPMorgan Emerging Local Markets Plus Index. “U.S. Treasurys” are defined as the
Barclays Capital U.S. Treasury Index; “U.S. Corporate Bonds” are defined as the Barclays
Capital U.S. Investment Grade Corporate Index; “U.S. High-Yield Bonds” are defined as
the Barclays Capital High Yield Corporate Index; “Developed-Market Equities” are
defined as the MSCI World Index; and “Emerging-Market Equities” are defined as the
MSCI Emerging Markets Index. All returns are shown in unhedged, U.S. dollar terms.
Country-Duration Contribution
Country-Duration Contribution
0.00
–0.05
–0.10
–0.15
–0.20
0.00
–0.05
–0.10
–0.15
–0.20
Japan
0.15
–0.16
Japan
–0.19
United
States Australia Mexico
0.44
–0.10
Australia
–0.03 –0.03
RAFI-Weighted Vs. Market
United
Kingdom
–0.07 –0.07
United
States
–0.06
South
Korea
–0.02
Mexico
–0.02
–0.05
Figure 7
Duration And Credit Spreads For Developed-Market Debt:
January 31, 2001, Through September 30, 2011
Option-Adjusted Spread (%)
0.60
0.50
0.40
0.30
0.20
0.10
0.00
-0.10
5.00
5.25
5.50
As of September 30, 2011
5.75 6.00 6.25 6.50 6.75 7.00
Duration (Years)
Source: Barclays Capital
Notes: Figure displays characteristics of the two different weighting methods for the
Barclays Capital Global Treasury Index.
policy is an important factor in sovereign risk and is not
captured by the relative size of a country’s GDP, landmass,
population or energy consumption.
In addition, the ability to repay debt is often not
as important as a sovereign government’s willingness
to repay its debt, and willingness can be even less
predictable. In 2008, Ecuador defaulted on its obligations,
despite having debt totaling 33 percent of GDP
Source: Vanguard, based on data from Barclays Capital and Arnott [2010]
Notes: Displays five largest (in absolute terms) countries’ duration contributions of
alternative weighting methodologies versus the cap-weighted developed-market
Treasury universe as defined by the Barclays Capital Global Treasury Index.
”Country-duration contribution” is computed as: (Weight Alternative − Weight Market cap) *
(Country Duration – Index Duration). In both alternative weighting methodologies, the
most positive country-duration contribution was less than 0.01 year. The cap-weighted
index duration was 6.52 years; GDP-weighted index duration was 6.12 years; and
RAFI-weighted index duration, 5.87 years. Barclays Capital data as of May 31, 2011;
RAFI weightings from Arnott [2010], applied to Barclays Capital country indexes.
in the prior year. 11 This was largely the result of a policy
decision, not any economic hardship; the decision to
default came after the election of a new president. In
contrast to these emerging market experiences, Japan
has managed a government debt load in excess of 100
percent of the country’s GDP for well over a decade,
owing to very low interest service on the debt and a
high domestic savings rate. Both country- and government-specific
factors determine the level of debt that is
sustainable for a given issuer.
Country-specific factors result in different governments
having very different risk profiles, regardless
of their debt levels. As shown in Figure 5, market participants’
past evaluation of emerging market countries’
debt has reflected the perceived risks of investing in
these nations. The country weights of alternative weighting
methods have the potential to increase risk meaningfully.
Japanese bonds and those of emerging markets are
not perfect substitutes: As shown in Figure 6, since the
1990s, emerging market debt issues have behaved more
like an equity investment, reflecting their risk. And,
although correlations will change over time, the low
14
January / February 2012

Figure 9 Figure 10
Alternative Index Weightings Result In Lower Duration And
Credit Quality In Emerging Market Sovereigns
Duration and credit spreads for emerging market debt:
January 31, 2009, through September 30, 2011
Option-Adjusted Spread (%)
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
3.0
3.5
4.0
As of September 30, 2011
5.0 5.5 6.0
Source: Barclays Capital
Notes: Displays the OAS and duration for the market-weighted and GDP-weighted
versions of the Barclays Capital Emerging Markets Local Currency Government
Universal Index from January 2009 through September 2011. We used the local
currency index to avoid any potential distortions from hard-currency issues.
4.5
Duration (Years)
Alternative Index Weights Shift Credit Risk Within
Emerging Market Sovereigns
Credit-quality distribution of alternative weighting methods in emerging market
sovereign debt, relative to market-cap weights
Weight Relative To Market
15
10
5
0
-5
-10
-15
-20
-1.6
Aa
7.5
-1.7
0.24
-18.3
A Baa Ba B
Source: Vanguard, based on data from Barclays Capital and Arnott [2010]
Notes: Displays distribution across Moody’s Investors Service credit-rating buckets of
alternative weighting methodologies versus the cap-weighted emerging market
sovereign universe, as defined by the Barclays Capital Emerging Market Government
Universal Government Index. Barclays Capital data as of May 31, 2011; RAFI
weightings from Arnott [2010], applied to Barclays Capital country indexes.
3.2
-0.3 -0.4
1.6
0.4
9.6
historical correlations for Japanese bonds suggest that
they may be a more effective diversifier than emergingmarket
bonds for a multi-asset-class portfolio.
This highlights our more general point: that focusing
solely on an issuer’s relative debt load ignores other factors
that may influence a government’s ability to repay its debt.
The governments of large, diverse, developed nations with
floating exchange rates and substantial economic, financial
and political infrastructure are very different entities
from governments of emerging markets, many of which
are less politically stable, with less credible institutions,
fixed exchange rates and concentrated, export-oriented
economies. We believe that these differences are evident to
bond market participants and incorporated into prices.
Alternative Weighting In
Developed Market Sovereign Bonds
Country-specific risk factors are likely the most
prominent drivers of an alternative-weighted bond
index’s performance. As shown in Figure 7, when
implemented within developed market sovereigns,
the country tilts that result from alternative weighting
methods do not seem to affect average credit quality of
the index, as measured by the option-adjusted spread
(OAS). 12 However, alternative weighting does lead to
lower overall index duration. 13 Figure 8 shows the largest
absolute country-duration contributions resulting
from the two alternative weighting methods examined,
relative to the cap-weighted market. Overweighting
a shorter-duration country (for example, the United
States) or underweighting a longer-duration country
(for example, Japan) both have the effect of lowering
the index average duration relative to the market. 14
The overall effect in both a GDP-weighted and RAFIweighted
index is to lower duration by about half a
year relative to the market. Lower-duration bonds, all
else equal, tend to offer lower yields, lower interest rate
risk and lower expected returns. Lower duration may or
may not be in line with an investor’s overall objective,
but it is important to note that reweighting countries
can change the interest rate sensitivity of the index.
Alternative Weighting In
Emerging Market Sovereign Bonds
In emerging market sovereigns, the results of alternative
weighting methods are slightly different, with
a decrease in duration but also an increase in credit
risk (Figure 9). Although both market-cap and GDP
weighting methods resulted in similar variability of
duration over time, the GDP-weighted index displayed
a much less consistent credit risk profile, as evidenced
by its large OAS dispersion. In addition to a general tilt
toward riskier issuers, the distribution across credit
qualities matters. Figure 10 illustrates that alternative
weighting methods can lead to concentrations within
certain risk buckets, potentially making it difficult to
assess an index’s overall credit quality. As with developed
sovereigns, the different risk characteristics of
alternative weighting methods within emerging sovereigns
may or may not be in line with an investor’s
overall objective. The important thing to note is that
reweighting countries can change both the credit risk
and interest rate sensitivity of the index.
Alternatively Weighted Indexes In
US Corporate Market
Similar to alternative weighting methods for sovereigns,
alternative indexes have been designed for U.S.
www.journalofindexes.com January / February 2012 15

Figure 11 Figure 12
Company Indebtedness And Bond Risk
Have A Weak Relationship
Nonmarket-Cap Weighting For Corporates
Underweights Financials
Credit spread and debt/equity ratio for S&P 500 companies: September 30, 2011
Option-Adjusted Spread (%)
8
7
6
5
4
3
2
1
0
0
2
4
R 2 = 0.0201
8 10
Source: Vanguard, based on data from FactSet and Barclays Capital
Notes: Includes S&P 500 companies that are also members of the Barclays Capital US
Investment Grade Corporate Index. “Option-adjusted spread” is the cap-weighted
average of each company’s outstanding debt issues. Debt/equity ratios shown are
from financial statements available as of October 25, 2011: For reference, the five
companies with the largest debt/equity ratios in decreasing order were Western
Union, Philip Morris International, Morgan Stanley, Goldman Sachs Group and
Marriott International. We excluded companies that fell into the Barclays Capital High
Yield Corporate Index, as they would have changed the scale of the chart, but these
companies exhibited a similar relationship, with an R-squared of 0.002. (”R-squared”
refers to the percentage of the variance in OAS that is explained by the variance of
issuers’ debt/equity ratios.)
corporate issuers, again with the objective of avoiding
the largest issuers. Alternative weights have largely
focused on metrics associated with issuer financial
performance, such as assets, revenue and cash flow.
Examples include the RAFI Investment Grade Bond
Index and the Barclays Capital Issuer Scored Corporate
Index for the investment-grade universe; and the RAFI
High Yield Bond Index for the high-yield universe. 15
Weighting methods based on financial-statement metrics
may differ slightly in implementation, but the general
framework is similar. Regardless of the specific
weighting criteria, these strategies all tend to redistribute
weight from larger issuers to smaller issuers, relative
to their cap-weighted counterparts.
As was true for the sovereign market, “large corporate
debt issuers” need not be construed here as having
any negative connotation: Prudent use of debt financing
can be a valuable tool for almost any corporation. 16
In addition, as we demonstrated within the sovereign
market, there are many factors to consider in bond risk
other than a company’s indebtedness. Figure 11 shows
the relationship between debt/equity ratios and yield
spreads over Treasurys for the debt-issuing companies
in the Standard & Poor’s 500 Index. The weak relationship
exhibited in the figure demonstrates that company
indebtedness has essentially no explanatory power with
regard to the risk of a company’s debt.
It should also be noted that corporations use debt
differently. Some companies explicitly use debt as part
of their business model: The classic banking model
involves borrowing on a short-term basis and lending
6
Debt/Equity Ratio
Sector distribution for weighting methods in US investment-grade
corporate market: As of June 30, 2011
Sector Weight
70%
60%
50%
40%
30%
20%
10%
0%
36%
33%
26%
Financials
53%
66%
Industrials
56%
11% 7%
11%
Utilities
■ Cap-Weighted ■ RAFI-Weighted ■ Issuer-Scored
Sources: Vanguard, based on data from Barclays Capital, Research Affiliates and SSgA
Notes: “Cap-weighted” is defined as the Barclays Capital US Investment Grade
Corporate Index; “RAFI-weighted” is defined as the RAFI Investment Grade Bond
Index; “Issuer-scored” is defined as the SPDRs CBND ETF (tracks the Barclays Capital
Issuer Scored Corporate Index).
at longer maturities, profiting from interest rate differentials.
Some companies use debt to a much lesser
degree (technology companies, for example), instead
preferring to issue equity and reinvest earnings to fund
operations. As a result, avoiding the larger issuers can
lead to sector tilts within the market. Figure 12, which
shows the sector distributions of the earlier-mentioned
RAFI Investment Grade Bond Index and the Barclays
Capital Issuer Scored Corporate Index, indicates that
alternative indexes tend to underweight the financial
sector, since banks are typically among the largest issuers
in the corporate market. 17 This sector tilt seems to
be a reaction to recent events in the market, as underweighting
financials would have resulted in outperformance
by alternatively weighted indexes over the past
several years. However, on a forward-looking basis, the
fact that the financial sector underperformed in the
past is not predictive of future performance.
Effects Of Basing Investor Decisions On Company Size
We next investigated the potential effects of basing
an investment decision solely on corporate bond
issuers’ size, by examining the distribution of interest
rate and credit risk across the U.S. market. Figure 13
displays duration and option-adjusted spreads across
different-sized U.S. investment-grade corporate issuers
over the past 10 years. As these figures show, risk
attributes change with the size of corporate bond issuers.
Larger issuers have typically been associated with
shorter-duration bonds than the market average. As
these largest issuers tend to be financials, the shorter
average duration seems consistent with the traditional
banking business model mentioned earlier. Smaller
issuers have been associated with higher spreads (and
higher default/credit risk). Considering that higher
yields are associated with higher credit risk and that
bond prices and yields move in opposite directions, it
16
January / February 2012

Figure 13
Interest Rate Risk And Credit Risk
Across The Corporate Market
US investment-grade corporate issuers: April 30, 2001− March 31, 2011
Duration (Years)
Option-Adjusted Spread (%)
7.0
6.5
6.0
5.5
5.0
4.5
4.0
2.5
2.0
1.5
1.0
0.5
0.0
5.4
Largest
20%
1.74
Largest
20%
5.9
20% -
40%
1.69
20% -
40%
Issuer Market-Cap Quintile
6.4 6.5
40% -
60%
40% -
60%
60% -
80%
Issuer Market-Cap Quintile
1.59
60% -
80%
6.0 6.0
Smallest
20%
Smallest
20%
Index
Index
Source: Vanguard, based on data from Barclays Capital
Notes: Figure is constructed from issuer-level statistics for constituents of the
Barclays Capital U.S. Investment Grade Corporate Index for April 30, 2001, through
March 31, 2011. Issuers were sorted by market cap in each month and then grouped
so that each quintile shown represented one-fifth of the market cap of the entire
index. Duration, OAS and Moody’s Investors Service rating are computed as the
market-cap-weighted average within each quintile over the entire time period.
makes sense that issuers with the smallest market cap
(and the lowest aggregate prices) tend to have higher
credit risk. This higher credit risk across issuer size also
contributes to the duration differences previously discussed,
as the smaller, lower-quality issuers typically
issue debt with higher coupon payments to compensate
issuers for their risk. 18
Investors considering a departure from a cap-weighted
corporate bond investment should note the sector
tilts associated with alternative indexes, as well as the
relationship between issuer size, duration and credit
risk across the entire market. Depending on one’s
specific objective, reweighting the market can lead to
risk-factor tilts that may make it difficult to evaluate an
index’s overall risk profile. For example, overweighting
the smallest issuers in the corporate market could
increase or decrease duration, depending on what
sector that extra weight is coming from. This strategy
could also increase the portfolio’s credit risk by overweighting
the issuers with higher spreads and lower
1.65
Aa A2 A3 A3
Moody’s Rating
2.00
A3/
Baa1
1.74
A2/
A3
credit scores. In this example, interest rate risk may
have increased or decreased, with lower credit quality,
relative to the overall market. The end result is an index
that is not necessarily “less risky” overall. As in the sovereign
market, focusing solely on issuer size in evaluating
ability to repay debt may overlook other issues that
market participants incorporate into prices.
Other Considerations In Fixed-Income Investing
As we have demonstrated, use of common alternative
index weighting methods introduces aspects
of active risk and redefines the market’s risk characteristics
where they are applied. These alternative
strategies use weighting factors that have no obvious
predictive power to explain the future risks or returns
of an issuer. The strategies merely reweight a market’s
constituents and involve tilts toward common
bond market risk factors, including country, sector,
duration and credit risk. In the process, alternative
weighting methods may make these important risk
considerations more difficult to interpret.
Given that the role of bonds in a portfolio has
traditionally been to provide price stability, income
and diversification versus more risky assets, it’s not
clear that risk-factor tilts are appropriate for most
fixed-income investors. Because we advocate a broadbased,
diversified approach to fixed-income investing,
we therefore believe market weights are an effective
way to gain exposure to the “consensus” portfolio.
However, we acknowledge that some investors may
wish to depart from market weights. Before doing so,
important factors should be considered, including the
risks, liquidity, turnover and implementation costs of
the proposed strategy. A key consideration to keep in
mind is that if an investor desires to over- or underweight
certain risk attributes relative to the market (by
targeting a certain country, a lower average duration or
a higher average credit quality, for example), such persistent
risk-factor bets may be achieved, and possibly
with great transparency and low cost, using traditional
cap-weighted indexes.
In the case of a global sovereign investment, various
country- or region-specific index products can be combined
to achieve consistent, targeted exposures to certain
countries, at the least allowing an investor to determine
his or her own allocation between developed and emerging
markets. Of course, this assumes that an allocation to
global fixed income as an asset class is appropriate for the
investor in question, which may or may not be the case.
For instance, the currency exposure inherent in international
bonds results in risk/return characteristics that
can be very different than those of an investor’s domestic
fixed-income market. 19 In addition, given emerging
market bonds’ high correlation with equity markets (as
in Figure 3), it’s not clear that this asset class performs
the traditional fixed-income function of diversifying an
equity position. Investors considering an overweight to
emerging-market bonds (or any allocation at all) might
www.journalofindexes.com January / February 2012 17

enefit from asking whether their objective could be
achieved with more traditional, risky asset classes such as
stocks or corporate bonds, since liquidity and implementation
costs are likely to be lower.
Investors within the corporate market should furthermore
consider whether a sector tilt is justified in their
portfolio, or merely a reaction to recent market events.
Underweighting a sector such as financials after it has
underperformed is a strategy that may not improve
of a particular market but, instead, a reweighting of
risks within that market. This introduces a component
of active risk into the index construction process,
whereby certain countries or corporations are
reweighted relative to the market’s consensus valuations
and the aggregate amounts already lent to these
issuers by investors. This is an important consideration,
since investors (in aggregate) incorporate all of the relevant
risk considerations of the issuer—including not
Alternative weighting methods systematically reweight issuers based
on predefined criteria that explicitly decouple an issuer’s weighting in an
index from the market cap of the issuer’s debt outstanding. Although
intuitively appealing, there is no obvious and consistent linkage
between debt outstanding and higher yields/lower prices.
future return expectations. In addition, if systematic riskfactor
tilts are in line with investors’ objectives, it may be
possible to change the risk profiles of these investments
through exposure to traditional segments of the bond
market—for example, by adding longer-duration and
lower-credit-quality bonds to a broader bond portfolio.
Not only do market-cap-weighted indexes and indexed
investments provide more reliable and transparent risk
exposure, they also tend to be among the least costly
ways to invest in fixed income. Particularly for assets
with lower expected returns, such as bonds, lower implementation
and management costs are important components—arguably
the most important components—in
an effort to improve a portfolio’s returns.
Conclusion
As a result of the global financial crisis of 2008-2009
and amid ongoing sovereign-debt concerns in much
of the developed world, investors have been searching
for investment strategies that might have mitigated
these crises, and in so doing, might have also outperformed
the cap-weighted market. With the benefit
of hindsight, some alternative weighting strategies
have seemingly been created to do just that, but their
ability to outperform in the future using these same
weighting strategies is uncertain.
We have shown that the common alternative index
strategies do not reflect the risk/return characteristics
only the issuer’s ability to repay the obligations but also
the issuer’s willingness to do so—into bond prices and,
therefore, into bond market capitalizations. Bonds
from any issuer, whether a country or a corporation,
cannot be issued without willing buyers at a price the
buyers deem appropriate to compensate them for the
risks they assume.
Alternative weighting methods, however, systematically
reweight issuers based on predefined criteria
that explicitly decouple an issuer’s weighting in an
index from the market cap of the issuer’s debt outstanding.
Although intuitively appealing, there is no
obvious and consistent linkage between debt outstanding
and higher yields/lower prices. On the contrary,
the U.S. and Japan are the world’s two largest
issuers of debt, and yet they benefit from yields that
are among the lowest in the world. On the other hand,
emerging countries tend to issue far less debt, but with
much higher yields. Although debt outstanding isn’t
an insignificant consideration when evaluating any
issue or issuer, we have shown that other factors are
more important to investors. We believe that marketcap-weighted
indexes best capture the broadest range
of risk factors that matter to, and are incorporated in,
prices set by bond investors.
The authors thank Didier Haenecour, John Malley
and Yan Pu, from the Vanguard Fixed Income and Risk
Management groups, for their excellent research assistance.
Disclosures
Notes about risk and performance data: All investing is subject to risk. Investments in bond funds are subject to interest rate, credit and inflation risk. Past performance is
not a guarantee of future results. The performance of an index is not an exact representation of any particular investment, as you cannot invest directly in an index. Current
and future portfolio holdings are subject to risk. Stocks of companies in emerging markets are generally more risky than stocks of companies in developed countries.
Foreign investing involves additional risks, including currency fluctuations and political uncertainty. Although U.S. Treasury or government agency securities provide
18
January / February 2012

substantial protection against credit risk, they do not protect investors against price changes due to changing interest rates. While the market values of government securities
are not guaranteed and may fluctuate, these securities are guaranteed as to the timely payment of principal and interest.
Endnotes
1. Market capitalization for a given fixed-income issuer is the sum of the value (measured in dollars, euro, yen, etc.) of all of that issuer’s outstanding debt securities. To
determine this issuer’s market-cap weight, one would divide that amount by the value of all the securities within the market being measured.
2. Another advantage of market-cap-weighting indexing relates to the zero-sum game, the idea that the asset-weighted overperformance in a market must equal the
asset-weighted underperformance in a market. (See Sharpe [1991] for a discussion, and Philips and Floyd [2010] and Philips [2011a] for an empirical evaluation of the
U.S. and offshore markets.)
3. Some index providers may adjust market weights to better reflect the investable opportunity set, accounting for things such as capital controls, liquidity and float (the
portion of an issue actually available for purchase). In our opinion, this paper’s discussion does not apply to these deviations from “pure” market weights, as they are
primarily done with investability and liquidity in mind, without attempting to deviate significantly from the market’s risk/return characteristics.
4. Investors with unique investment objectives may deviate from the broad bond market to better form a portfolio that fits their own circumstances. For example, pension
funds often manage their assets to a predefined liability through the use of longer-duration bonds [Stockton, Donaldson and Shtekhman, 2008]. In such cases,
the investor’s initial market is defined differently so as to isolate securities with specific risk/return attributes (a specific beta).
5. See the Barclays Capital website for specific details on these indexes’ construction.
6. See Arnott et al. [2010] and Arnott [2010]. We refer to this method as “RAFI-weighted,” based on the notation in these publications.
7. As suggested by the low R-square of the regressions. R-squared is a measure of how much of a portfolio’s performance can be explained by the returns from the overall
market (or a benchmark index).
8. See Davis et al. [2010] for a discussion of the drivers of interest rates in the United States.
9. See Reinhart and Rogoff [2009] for a discussion of this and other similar examples.
10. See Krugman [2000] for a review of several currency crises.
11. Source: IMF.
12. Option-adjusted spread refers to the yield spread of a bond versus that of a comparable-maturity U.S. Treasury bond, with an adjustment for the value of any embedded
options.
13. Duration is a common measure of the sensitivity of the price of a bond to a change in interest rates.
14. Although a RAFI weighting methodology would underweight the U.S. when implemented within a global investment set, the weighting method would overweight the
U.S. when implemented within a developed-market investment set. In a global set, alternative weighting methods overweight emerging market countries, with the
majority of the weighting difference being shifted to these emerging countries by Japan and other G-7 governments, not the U.S.
15. For a discussion of the backtested performance of the RAFI weighting methodology, see Arnott et al. [2010]. For index construction details and fact sheets, see the
Research Affiliates and Barclays Capital websites: http://www.researchaffiliates.com/rafi/index.htm and http://us.ishares.com/home.htm, respectively.
16. Debt issuance can function as part of an effective corporate governance program, restricting cash that might otherwise be used by corporate management on risky or
frivolous projects. Issuing debt avoids diluting the ownership stake of equity investors. Also, since interest service is expensed and offsets taxable income in the United
States, debt issuance provides a tax benefit to issuers.
17. As of June 30, 2011, the largest market-cap quintile of the investment-grade corporate market comprised 70 percent financial issuers, while financial issuers made up
only 36 percent of the total market.
18. Higher yields are associated with lower durations, all else equal.
19. See “Global Fixed Income: Considerations for U.S. investors” [Philips et al. 2011b], for a discussion of the merits of international fixed income from the U.S. perspective,
with a particular focus on the challenges that currency presents for this asset class.
References
Arnott, Robert, 2010. “Debt Be Not Proud.” Journal of Indexes (Nov/Dec.): pp.10-17; available at IndexUniverse.com.
Arnott, Robert, Jason Hsu, Feifei Li and Shane Shepherd, 2010. “Valuation Indifferent Weighting for Bonds.” Journal of Portfolio Management 36(3): pp. 117-30.
Brinson, Gary, L. Randolph Hood and Gilbert Beebower, 1986. “Determinants of Portfolio Performance.” Financial Analysts Journal 42(4): pp. 39-48; reprinted in Financial
Analysts Journal (50th Anniversary Issue) 51(1): pp. 133-38.
Davis, Joseph, Roger Aliaga-Díaz, Donald Bennyhoff, Andrew Patterson and Yan Zilbering, 2010. “Deficits, the Fed, and Rising Interest Rates: Implications and
considerations for bond investors.” Valley Forge, Pa.: The Vanguard Group.
Krugman, Paul, 2000. “Currency Crises.” Chicago: University of Chicago Press.
Philips, Christopher and Sarah Floyd 2010. “The Case for Indexing: European- and Offshore-Domiciled Funds.” Valley Forge, Pa.: The Vanguard Group.
Philips, Christopher, 2011a. “The Case for Indexing.” Valley Forge, Pa.: The Vanguard Group.
Philips, Christopher, Joseph Davis, Andrew Patterson and Charles Thomas, 2011b. “Global fixed income: Considerations for U.S. investors.” Valley Forge, Pa.: The
Vanguard Group.
Philips, Christopher, Francis Kinniry Jr., David Walker and Charles Thomas, 2011c, forthcoming. “A review of Alternative Approaches to Equity Indexing.” Valley Forge,
Pa.: The Vanguard Group.
Reinhart, Carmen and Kenneth Rogoff, 2009. “This Time Is Different: Eight Centuries of Financial Folly.” Princeton, N.J.: Princeton University Press.
Sharpe, William, 1991. “The Arithmetic of Active Management.” Financial Analysts Journal 47(1): pp. 7-9.
Stockton, Kimberly, Scott Donaldson and Anatoly Shtekhman, 2008. “Liability-Driven Investing: A Tool for Managing Pension Plan Funding Volatility.” Valley Forge, Pa.:
The Vanguard Group.
www.journalofindexes.com January / February 2012 19

A Fundamentally Weighted
Broad-Based Fixed-Income Index
An alternative bond investing paradigm
By Shane Shepherd
20
January / February 2012

At the bare minimum, a benchmark index should
measure the performance of an investment strategy
or asset class. But as far as investors are concerned,
although most fixed-income indexes meet this threadbare
requirement, they fail miserably on most other counts. For
starters, cap-weighted fixed-income indexes deliver suboptimal
portfolio allocations, impairing performance. Many bond
indexes also fail to provide adequate investability and liquidity.
In addition, many bond indexes also show extremely high
turnover, lack of adequate diversification and unstable risk
characteristics. Finally, it is difficult to properly determine
the weight of many issues in a given index because bonds are
traded over the counter, oftentimes at infrequent intervals.
Thus a key variable for the index weight—the price of the
bond—is not known with precision.
In this paper we show that, in the presence of real-world
frictions and imperfect capital markets, a cap-weighted aggregate
global bond portfolio leads to suboptimal allocations and
performance. In contrast, a simulated global broad bond index
that is fundamentally weighted (“fundamentally weighted
portfolio”) and that selects in a liquid, representative fashion,
creates superior performance over time, based on simulated
performance data since 1997. The six components of our
broad bond index all show outperformance in the range of 75
bps to over 400 bps annually, and the broad portfolio provides
110 bps outperformance over a cap-weighted benchmark of
the same bonds, and 161 bps annual outperformance over the
Barclays Capital Global Aggregate Index. 1 Additionally, these
fundamentally weighted portfolios provide improved liquidity
and investability, thus representing achievable returns.
Theoretical Overview
If the bond markets are less than perfectly efficient, the
resulting mispricings will result in a return drag embedded in
the cap-weighted bond indexes. This phenomenon has been
shown in the equity markets by Arnott, Hsu and Moore (2005);
Hsu (2006); and in the fixed-income markets by Arnott, Hsu, Li
and Shepherd (2010). The optimality of cap-weighted indexes
as investment options depends crucially on the assumptions
of perfectly efficient capital markets and rational investors
with mean-variance utility preferences. As we begin to weaken
these assumptions, market mispricings enter into the prices we
observe in the marketplace. Because a cap-weighted portfolio
bases its constituent weights on price, there will be a systematic
correlation between pricing errors and portfolio weights
that leads to suboptimal allocations and performance.
Additionally, traditional fixed-income indexes are exposed
to the “bums” problem, as noted by Enderle, Pope and Siegel
(2003). As a corporation or country issues more and more
debt, it becomes a bigger and bigger part of the index. Because
traditional indexes are weighted by size of issuance, these
heavily indebted “bums” will dominate the index weights.
From a benchmarking standpoint, that may not be a concern:
The indexes are, after all, reflecting the current opportunity
set. However, from the investor’s perspective, why should we
allocate a higher portion of our portfolio to big debtors just
because they wish to borrow more?
We view uncertainty about default risk and inflation rates
as the primary source of market inefficiencies. 2 If market participants
have properly estimated the amount of default risk to
which they are exposed by holding a bond, then the current
yield exactly compensates them for holding that risk. However,
if they underestimate the default risk, then they end up holding
a bond with too low of a yield (and, thus, too high of a price); if
they overestimate the risk of default, they hold a bond with an
overly high yield and too low of a price. Now we see how the
return drag enters into standard indexes.
Because returns are calculated based on the relative dollar
value of each holding, those bonds that trade too rich will
have a higher price and, thus, a higher weight, while those
bonds that trade cheaply will have a lower price and, therefore,
a lower weight. The cap-weighted index will have a
perfect correlation between the pricing errors and its portfolio
weights, systematically overweighting overvalued bonds and
underweighting undervalued bonds. The index is overexposed
to bonds where the default risk outweighs the yield, and
underexposed to bonds where the premium yield exceeds the
default risk. We may not know which bonds these are a priori,
but if these mispriced bonds exist in the marketplace, this
systematic correlation will always hold.
Indexing By Fundamentals
We demonstrate that this return drag can be easily sidestepped
by creating an index with portfolio weights that do not
rely on market prices. By weighting according to fundamental
variables, those companies or countries with better economic
fundamentals receive higher weights and those with weaker
fundamentals receive lower weights. In a sense, this results in
weighting according to the issuer’s ability to make payment
on its debt rather than on the amount of debt outstanding.
Additionally, a systematic rebalance back to these fundamental
weights results in contratrading against market price movements.
This contratrading results in a systematic buy-low/sellhigh
strategy that contributes to the long-run outperformance.
Shepherd (2011) shows that this process leads to a price return
component that accounts for the outperformance of fundamentally
weighted corporate bond indexes with equivalent yields.
A higher-yield component would suggest that the outperformance
was being driven by increased allocation to risk factors.
Fundamentally weighted bond indexes are constructed
in six distinct areas: developed markets sovereign debt;
emerging markets sovereign debt (denominated in local
currency); emerging markets sovereign debt (denominated
in USD); global developed markets investment-grade
corporate debt; global developed markets high-yield corporate
debt; and emerging markets corporate debt.
The weighting process detailed in Arnott et al. (2010) is
followed. To determine portfolio weights for the corporate
bond segments, we use a combination of four headline
accounting numbers: sales, cash flow, dividends paid and
book value of assets. We measure average five-year numbers
for the first three factors and use the most recent book value
of assets. The fundamental weight for each company is then
calculated as the average of each of the four fundamental
scores. This averaging leads to a more robust, slowly moving
measure of economic size for each company. We then select
www.journalofindexes.com January / February 2012
21

which bonds to include in the index from each company.
Up to three bonds are selected for each corporation, pulling
the largest and most liquid bond from each part of the yield
curve at which that corporation has eligible debt issues. The
portfolios are rebalanced annually back to target weights
and reviewed monthly for inclusion based on index rules.
To determine the weighting for the country segments,
we use a combination of gross domestic product (GDP);
population; land area (linearized by square root); and energy
consumption. These four factors represent a country’s
capital, labor force, natural resources and technological
sophistication, which all contribute as primary inputs to an
economy’s growth. We use five-year averages for each factor
and compute a composite fundamental weight as the equally
weighted average of these four factors. To provide better
diversification, the country weights are then rescaled such
that the weights are shrunk toward a more equally weighted
portfolio. For the issue representation, we select up to three
bonds from each country, taking the most liquid bond from
each part of the yield curve. The country portfolios are also
rebalanced annually back to target portfolio weights and
reviewed monthly for index rules.
We then form the fundamentally weighted portfolio by
pulling all of these component parts together and weighting
each sector by the face value of that sector. Because
the corporate accounting variables and the country economic
variables produce varying fundamental weights,
there is no direct way to weight the sectors by fundamentals.
By using face value of debt, we provide a broad measure
of size that still allows an annual sector rebalance
back to a slowly moving target weight (in effect, trading
against movements in credit spreads) and maintains low
tracking error to cap-weighted benchmarks.
Performance And Characteristics
We calculate monthly returns in unhedged USD for each of
these indexes as far back as constituent data is available. The
time period is from January 1997 through June 2011, except
for the global high yield (where the start date is January 1998)
and emerging markets sovereign local currency (start date
of January 2001). We compare our results to both the published
benchmark index from which we gather constituents
(named in each table) as well as a cap-weighted universe of
those bonds that are included in the fundamentally weighted
index. Not all bonds from the published benchmarks can be
matched to fundamental variables. 3 Additionally, our data
do not fully capture intramonth changes in the benchmark
index. By calculating returns for the cap-weighted portfolios,
we remove any questions about whether these discrepancies
are driving the outperformance numbers.
Each of the six component fundamentally weighted portfolios
outperform both the cap-weighted portfolio benchmark
and the relevant published index, as shown in Figures
1 and 2. The amount generally ranges from 71 bps to 182
bps, except for the Global High Yield Corporate Bond Index,
which outperforms by 473 bps annually. The volatility for the
fundamentally weighted portfolios is typically comparable
or lower than the benchmarks, except for the Developed
Sovereign Index. In all cases, weighting according to fundamentals
has resulted in a significant improvement to the
Figure 1
Fundamentally Weighted Portfolios Vs. Cap-Weighted Benchmarks
Portfolio
Return
Standard
Deviation
Sharpe
Ratio
Average
Credit Rating
Duration
Excess
Return
Fundamentally Weighted
Developed Sovereign Index
Developed Sovereign
Cap-Weight Index
BofA Merrill Lynch
Global Government Index
Fundamentally Weighted EM
Sovereign Local Currency Index
EM Sovereign Local
Currency Cap-Weight Index
JP Morgan GBI-EM
Global Diversified Index
Fundamentally Weighted
EM Sovereign USD Index
EM Sovereign USD
Cap-Weight Index
BofA ML USD Emrg Mkts
Sovereign Plus Index
6.84 7.76 0.51 AAA/AA1 5.83 0.82
6.03 7.18 0.44 AA1 5.76 0.01
6.02 7.05 0.44 – – –
13.54 10.35 1.11 BBB+ 3.86 1.32
12.37 11.61 0.89 A- 4.12 0.14
12.23 11.59 0.88 – – –
12.06 11.41 0.80 BB2 5.48 1.67
10.87 12.93 0.62 BB2/BB3 5.97 0.49
10.38 13.25 0.57 – – –
Source: Research Affiliates, based on data from Bloomberg. Developed Sovereign and EM Sovereign USD portfolios cover the time period January 1997–June 2011;
EM Sovereign Local Currency portfolios cover the period January 2001–June 2011.
22
January / February 2012

Figure 2
Fundamentally Weighted Corporate Bond Indexes Vs. Cap-Weighted Benchmarks
Portfolio
Return
Standard
Deviation
Sharpe
Ratio
Average
Credit Rating
Duration
Excess
Return
Fundamentally Weighted
Glb Inv Gr Corp Index
Investment Grade Corporate
Cap Weight Portfolio
BofA Merrill Lynch
Global Corporate Index
Fundamentally Weighted Global
High Yield Corporate Bond Index
Global High
Yield Cap Weight
BofA Merrill Lynch
Global High Yield Index
Fundamentally Weighted
Emrg Mkts Corp USD Index
EM Corporate
Cap-Weight Index
BofA Merrill Lynch US Emerging
Markets Credit Plus Index
7.16 6.57 0.65 A1/A2 5.28 0.74
6.62 6.47 0.58 A2 5.37 0.19
6.43 6.53 0.54 – – –
11.77 11.24 0.80 BB3/B1 4.40 4.73
7.75 9.65 0.52 BB3/B1 4.43 0.71
7.04 10.75 0.40 – – –
11.54 9.08 0.99 BB1/BB2 4.55 1.82
9.90 9.60 0.77 BB1 4.84 0.05
9.71 10.01 0.71 – – –
Source: Research Affiliates, based on data from Bloomberg. The Global Investment Grade Corporate and Global High Yield Corporate portfolios cover the time period January
1998–June 2011; the EM Corporate portfolios cover the period January 1999–June 2011.
Sharpe ratio. Finally, it is not just the composite indexes that
outperform; each of the four component factors deliver outperformance
over the benchmarks for all six indexes. 4 It is
not crucial which fundamental measures of size are used; as
long as there is no correlation with market prices and pricing
errors, we are able to remove the return drag.
Additionally, we can see that the outperformance is not
due to increased exposure to duration or credit risk. The
fundamentally weighted portfolios have in-line or lower
average duration than the benchmarks for each portfolio.
Average credit ratings are typically in line or half a grade
better than the benchmark credit risk. This tendency makes
intuitive sense. We expect that companies or countries
with greater amounts of debt relative to their fundamental
size generally receive poorer credit ratings, and those companies
will typically have higher weights in a cap-weighted
index than in a fundamentally weighted index.
Comparing the relative outperformance with the anticipated
level of inefficiency in each market provides additional
support for the “return drag” hypothesis. If the
performance differential is generated by systematic mispricing
effects, we would expect that return drag to be
greater as the level of inefficiency grows. This is indeed the
result. We see that the emerging market sovereign debt
portfolios, with greater volatility and inefficiencies than the
developed sovereign portfolios, provide 1.5 to 2 times the
amount of outperformance. We see that the investmentgrade
portfolios provide less outperformance than either
the Emerging Markets Corporate Index or the Global High
Yield Corporate Index. Greater market inefficiencies lead
to larger return drags and greater ability to outperform the
increasingly suboptimal cap-weighted index.
Figure 3 shows the results for the broad-based fundamentally
weighted portfolio compared with our internally generated
cap-weighted benchmark as well as the BarCap Global
Aggregate. The fundamentally weighted outperformance of
1.61 percent annually comes, much like the component
pieces, without assuming additional credit or duration risk.
The broad-based fundamentally weighted portfolio shows a
superior credit rating by about half a grade, and a lower average
duration by 0.11 years. By comparing the cap-weighted
benchmark returns to the BarCap Aggregate returns, we see
that about 50 bps of the outperformance over the BarCap
Aggregate is due to the selection effect of excluding some sectors
of the debt markets (or perhaps from minor differences
between the BarCap Global Aggregate and the BofA Merrill
Lynch component indexes). Figure 4 shows the average sector
allocations, with the relatively more efficient developed sovereign
and global investment-grade corporate pieces dominating
the aggregate allocation for both the fundamentally
weighted sovereign and the cap-weighted portfolio. 5
Given the higher weights of developed market debt, the
1.11 percent annualized outperformance versus the capweighted
benchmark is quite impressive: the developed
sovereign portfolio’s excess return of 0.82 percent and the
investment-grade corporate credit portfolio’s outperformance
of 0.74 percent show that the annual reweighting of
the fundamentally weighted index contributes a substantial
proportion of the outperformance. Even though these two
continued on page 61
www.journalofindexes.com
January / February 2012
23

Market-Implied Default
Probabilities And Credit Indexes
Weighting issuers in bond indexes by risk of default
Terry Benzschawel, Cheng-Yen Lee, Brent Hawker and David Craft
24 January / February 2012

We describe a method for estimating default
probabilities (PDs) for risky obligors and
apply that method to generate PDs for major
sovereign issuers. We then show how these marketimplied
PDs can be used to weight obligor contributions
to indexes of risky credits and highlight some
features of PD-weighted indexes. Although several useful
approaches for estimating PDs exist, including fundamental
analysis, statistical models, structural models
and the risk-neutral approach, none has proven entirely
satisfactory. This is particularly true for sovereign credits
(which often default strategically), heavily leveraged
financial firms, private firms whose financials are not
readily available and municipal bond issuers. Interest
in a generally acceptable method for estimating PDs
has increased recently due to the requirements of the
impending Basel regulations, as well as firms’ needs to
assess counterparty credit risk and make credit value
adjustments to profit-and-loss on trading positions.
Another potential use for market-implied PDs is for
differential weighting of obligor contributions to indexes
of default-risky assets. Financial indexes offer investors
a diversified exposure to given market segments as
well as performance benchmarks for financial managers.
However, monthly index adjustments can introduce large
discrete changes in composition, requiring costly buying
and selling of securities. This can be particularly severe if a
large issuer’s credit deteriorates, resulting in exclusion. We
introduce in this report a method for constructing indexes
weighted by PDs from the market-implied PD model.
Although we describe the method of constructing the index
in terms of Citigroup’s World Government Bond Index, 1
the method is general enough to be applied to any index
with a credit-based criterion for inclusion.
Market-Implied PDs
We developed a procedure for measuring expectations
of obligors’ PDs from the credit spreads and spread
volatilities of their assets. The method first uses credit
spreads, along with obligors’ model-based PDs, to estimate
the current credit risk premium in order to determine
the spread compensation per unit of default probability.
Then, any firm’s bond spread and spread volatility
are measured with respect to the current risk premium
and, given an assumed value of recovery in default, can
be used to infer a market-implied PD.
The market-implied PD framework views obligors’
spread as consisting of two parts: compensation for default;
and compensation for credit spread volatility. The method
is based on demonstrations that credit spreads, on average,
are linear functions of spread volatility on logarithmic
axes. That linear relationship holds, in general, for spreads
overall and for spreads of bonds within individual ratings
categories. Also, we assume that investors require the same
level of spread compensation per unit of spread volatility
regardless of its source. We go into more depth on the
mechanics of this measure below.
Although there is demand for a market-based PD measure,
accurate estimation of PDs has proven difficult for
several reasons. First, changes in agency ratings tend to lag
market spreads (Figure 1). That is, credit rating changes
trail market perceptions of credit quality. Also, changes in
default rates and credit spreads vary over the credit cycle
and, while related, are not perfectly correlated (Figure 2).
A major difficulty in inferring PDs from bond prices is that
bond spreads contain a risk premium that is not related to
default. If so, it should be possible to decompose a credit
spread into components such that s = s d
+ s λ
, where s d
is the
compensation for default and s λ
is the credit risk premium.
For a bond of duration T, the spread compensation for
default, s d
, can be approximated as:
where LGD is loss-given-default or 1-R, where R is the
recovery value in default. The value of s d
can be thought of
as the amount of spread necessary to equal the expected
return of a U.S. Treasury issue of similar duration for a
given expected default and recovery for the risky bond.
Also, since s = s d
+ s λ
, if the credit spreads are used along
with estimates of default probabilities, one can solve for the
risk premium, defined as s λ
, using the following relation:
Equations 1 and 2 can be derived directly from the bond
price versus yield relationship. 2
Figure 1
Deviation From Spread Of
Target Rating (bp)
15
10
5
0
(5)
(10)
(15)
(20)
(25)
Credit Spreads Relative To Monthly
Average Of Destination Ratings*
Upgrades 1985-1999*
Spreads Tighten
Downgrades
Spreads Widen
(30)
(36) (30) (24) (18) (12) (6) 0 6 12
Months From Ratings Change
Month of
Upgrade/
Downgrade
*The sample contained over 2,500 downgraded bonds and over 2,000 upgrades
(from Benzschawel and Adler [2002]).
Source: Citi Investment Research & Analysis
Given estimates of average PDs and recovery values by
rating category, one can calculate the spread compensation
for default using Equation 1 and the residual nondefault
spread implied by Equation 2. These calculations
were performed on monthly credit spreads from 1994-2010
using corporate bonds in Citigroup’s U.S. Broad Investment
Grade (BIG) Index and Citigroup’s High Yield Market
Index [Citigroup, 2011], with a value of T = 4.5, roughly the
average duration of the bonds in question, and assuming R
= 40 percent of face value (Figure 3). Notice that for all rat-
(1)
(2)
www.journalofindexes.com January / February 2012
25

Figure 2
Mcap/GDP
14%
12%
10%
8%
6%
4%
2%
Annual Average High Yield Credit Spreads
and Annual Default Rates (1986-2010)
High Yield
Spreads
Default Rates
0%
1985 1988 1991 1994 1997 2000 2003 2006 2009
Sources: Standard & Poor’s and Citi Investment Research & Analysis
1600
1400
1200
1000
ing categories, the breakeven compensation for default is
only a small fraction of the overall spread compensation. In
fact, nearly all of the spread for any investment-grade bond
(rated triple-B-minus or better) is due to the credit risk
premium, rather than compensation for default. For high
yield, default plays a greater relative role, but the average
risk premium can still be as large as 700 bp.
We have shown that the credit risk premium varies
over time. 3 That is, the risk premium for investment-grade
bonds averages about 125 bp, but has been below 50 bp
during periods of high liquidity in the mid-1990s and mid-
2000s, and has risen as high as 650 bp during the credit/
liquidity crisis of 2009.
Although the compensation for default is specified in
Equation 1, solving for the risk premium s λ
in Equation 2
assumes that investors require the same level of spread
compensation per unit of spread volatility, regardless of its
source. That assumption is supported by the observations
of monthly spreads and spread volatilities from 1994 to
mid-2010 presented in Figure 4. That is:
• Logarithms of averages of five-year rolling spread volatilities
increase linearly as credit rating deteriorates
(blue symbols in left panel)
• For all rating categories, spread volatilities are
roughly one-third of average credit spreads (red
symbols in left panel)
• The variability of the volatility-to-spread ratios is
also constant across rating categories (green symbols
in left panel)
• Average five-year trailing spreads, grouped by rating
categories, are linear functions of their five-year volatilities
(middle panel)
800
600
400
200
0
High Yield Index Spread (bp)
• Logarithms of average spreads by rating category after
subtraction of spread due to default s d
using median
PDs from Sobehart and Keenan’s [2002, 2003] HPD
model using Equation 1 (right panel) are linear functions
of log spread volatilities
The results in Figure 4 suggest that, on average, the market
charges the same amount of spread per unit of volatility
regardless of whether it comes from a high-quality credit or
a risky one. Accordingly, the hypothesis is that for all credits
at a given time t, s λ
in Equation 2 can be expressed as:
where s λ
is the spread compensation over and above that
for default, σ is the volatility of that spread and λ t
is the
market price of risk at time t. Using Equation 3, one can
rewrite Equation 2 as:
For Equation 4 to be useful, the long-term relationships
between spreads and spread volatilities in Figure 4 must
hold over a much shorter time horizon. Also, estimating
the daily risk premium λ poses challenges. That is, even if
one has bond spreads and volatilities, and can assume a
reasonable value for LGD (e.g., 1-R = 60 percent), Equation
4 requires estimates of default probabilities p T
in order to
solve for λ. Yet it is precisely those probabilities that the
model is trying to estimate.
The rationale behind our approach to estimating the
daily risk premium is described in detail elsewhere 4 and is
described only briefly here. To calculate the risk premium,
we found it necessary to first convert each obligor’s set of
credit spreads to a single equivalent spread for a maturity
of one year. 5 To do this, term structures of firms’ bond
spreads were fit to curves using a Nelson-Siegel [1987] procedure.
If no term structure is available for a given firm, the
model uses an average term structure derived from spreads
of firms having similar PDs.
Using historical average PDs from the rating agencies
to generate nondefault spreads for estimating the
risk premium proved problematic. Since agency default
rates increase logarithmically with decreasing credit
quality, mean default rates are poor estimates of central
tendency. Rating category median default rates might
be used to establish monthly norms, but this is problematic
because median values for PDs would remain
(3)
(4)
Figure 3
Trading
Symbol
Average Default Probabilities, Credit Spreads, Spreads Due To Default,
And Credit Risk Premiums (Non-Default Spreads) By Credit Rating Category (1994-2010)
AAA AA+ AA AA- A+ A A- BBB+ 3-Month BBBAverage
BBB- BB+ BB BB- B+ B B- CCC+
Daily Volume
4.5 Year Implied PD 0.3% 0.3% 0.4% 0.5% 0.6% 0.7% 0.9% 1.0% 1.3% 1.6% 1.9% 2.3% 2.7% 3.8% 5.7% 8.2% 11.8%
Average Spread (bp) 70 87 88 104 113 122 137 160 178 223 326 366 383 438 511 588 891
Spread Due To Default (bp) 1 4 5 6 8 9 11 14 17 21 25 30 36 51 77 113 164
Non-Default Spread (bp) 69 82 83 98 10 112 126 146 161 201 301 336 347 386 433 476 727
Sources: Standard & Poor’s and Citi Investment Research & Analysis
26 January / February 2012

Figure 4
A: Average Of Rolling 5-Year Averages Of Spread Volatilities By Rating
B: Average 5-Year Trailing Spreads Vs. 5-Year Trailing Spread Volatilities
C: Non-Default Credit Spreads By Rating Vs. Spread Volatility For June 2010
A B C
1000
0.6
3163
1000
Spread Volatility (Left Axis)
Vol / Spread Ratio (Right Axis)
0.5
R 2 =0.98
StdDev of Vol / Spread Ratio
316
1000
0.4
316
Spread Volatility (bp)
100
32
10
AAA
AA+
AA
AA-
A+
1994-2010
A
A-
BBB+
BBB
BBB-
BB+
BB
BB-
B+
B
B-
CCC+
0.3
0.2
0.1
0.0
Ratio
5-Year Average Spread (bp)
316
100
1994-2010
32
32 100 316 1000
5-Year Trailing Spread Vol (bp)
Non-Default Spread (bp)
100
R 2 =0.97
36
3 10 32 100 316
Spread Volatility (bp)
Source: Citi Investment Research & Analysis
Figure 5
A: Estimating The Value Of The Risk Premium λ τ
From One-Year Non-Default Spreads
B: Bond Spreads Vs. Spread Volatilities For August 1, 2011
C: One-Year Equivalent Spreads Vs. Volatility And Lines Of Constant PDs Corresponding To Major Letter Rating Boundaries
A 300
B 10000
Spread
C
1
s = λσ
3300
3 = λσ -{ In[1 - (pT • LGD)]
T }
100
1000
1-Yr Non-Default Spread (bp)
30
10
3 10 30
Slope = 1
Spread Volatility (bp)
100 300
Equivalent 1-Yr Credit Spread (bp)
333
100
33
Spread Volatility (bp)
Non-Default
Spread,
λσ
10
1 3 10 33 100 333 1000
1-Yr Spread (bp)
10000
3000
1000
300
100
30
10
3
1
CCC
B
BB
BBB
A
AA
AAA
CC
Cheap
Region
.02%
C
.05%
0.1%
0.3%
.62%
1.4%
4.2%
Spread Volatility (bp)
26%
Rich
Region
64%
Fair Value Line
Lines of
Constant PD
Source: Citi Investment Research & Analysis
constant over the credit cycle, and median default rates
are zero for all investment-grade rating categories. For
these reasons, PDs for commercial and industrial firms
from Sobehart and Keenan’s HPD model were used to
calculate the nondefault spreads necessary to estimate
the daily credit risk premium. Thus, PDs from the HPD
model, where available, are used as “initial guesses” of
firms’ market-implied PDs to enable estimation of the
risk premium using Equation 4. 6
The left panel of Figure 5 shows a fit of one-year
equivalent nondefault spreads from a sample of several
thousand firms for June 30, 2010. To construct that plot,
default spreads (s d
in Equation 1) were subtracted from
firms’ one-year spreads and fit with a line of slope equal
to one in accordance with the assumption of a constant
credit risk premium. The y-intercept of that line is the log
of the daily risk premium, λ t
. Given the risk premium, one
can estimate PDs as demonstrated in the middle and right
panels of Figure 5. The upward-sloping straight line in the
middle plot is the line of λ t
times σ; spreads falling on this
line have an implied one-year PD, p 1
, equal to zero. The
curved line is fit to log spreads versus log volatilities. Then
firms’ market-implied PDs are inferred from an orthogonal
vector projection to the curve line as shown for a sample
point by the dashed lines in the figure. The spread and
spread volatility implied by a firm’s vector projection are
used with Equation 5 and an assumed LGD to solve for that
firm’s default probability.
Similar vector projections to that shown in the
middle panel of Figure 5 can be applied to any firm
whose spread and spread volatility are known. For
example, firms’ one-year spreads and volatilities for
August 1, 2011 appear in the right panel of Figure 5.
The lines with negative slopes that have PD labels are
lines of “iso-PDs”; all points along each line would be
referenced to the same p 1
on the best-fit line through
the data points. The particular lines shown correspond
to implied rating category boundaries. 7 The rating
category boundaries are determined by matching the
frequencies of points in each rating category with the
frequencies of firms in each category in our sample.
For example, if there are 10 triple-A firms in our uni-
www.journalofindexes.com January / February 2012
27

verse, the 10 firms with the lowest default probabilities
are assigned to that category. The next group of firms
ranked by PDs is assigned a rating of double-A-plus,
and so on. The median market-implied default probability
is calculated for each rating category, and boundaries
between adjacent categories are computed as the
geometric means of these medians.
Generating PDs For Global Sovereign Issuers
As a preliminary step in constructing a PD-weighted
index of government bonds, we show how the model can
be used to estimate PDs for major sovereign bond issuers.
Although generating accurate estimates of PDs for sovereigns
has proven difficult [Wirz, 2011], one can estimate
sovereign PDs using the market-implied method. Spreads
Figure 6
Values Of Market-Implied PDs And Five-Year CDS Spreads For Sovereign Debt Issuers, August 1, 2011
Issuers Listed In Order Of Default Risk
Country
Market Implied
PD(%)
5-Yr CDS
Spread (bp)
Country
Market Implied
PD(%)
5-Yr CDS
Spread (bp)
Greece 38.82 2020
Ireland 13.91 758
Portugal 13.78 763
Belarus 12.32 –
Venezuela 11.62 990
Pakistan 10.50 924
Ecuador 9.28 378
Argentina 4.65 586
Dominican Republic 3.16 378
Serbia 2.94 353
Ghana 2.89 –
Georgia 2.60 –
Vietnam 2.59 326
Jordan 2.57 338
Nigeria 2.21 99
Gabonese Republic 1.96 –
Bahrain 1.91 230
Romania 1.70 231
Morocco 1.65 169
Croatia 1.58 266
El Salvador 1.26 336
Hungary 0.69 288
Bulgaria 0.64 203
Turkey 0.63 156
Peru 0.57 –
Lithuania 0.55 192
Spain 0.54 244
Lebanon 0.48 351
Qatar 0.36 99
Russia 0.35 140
Poland 0.34 149
Italy 0.33 162
Panama 0.29 –
South Africa 0.29 125
Chile 0.25 75
Czech Republic 0.22 83
5-Year CDS Spread (bp)
Colombia 0.20 109
Philippines 0.20 137
Uruguay 0.17 165
Brazil 0.14 110
Hong Kong 0.13 –
Mexico 0.08 102
Belgium 0.07 130
Korea 0.04 104
Norway 0.04 18
New Zealand 0.04 67
Germany 0.03 34
Israel 0.02 119
Switzerland 0.02 35
Australia 0.01 57
United Kingdom 0.01 –
1000
560
300
Denmark 0.01 –
France 0.01 71
Finland 0.01 32
Netherlands 0.01 33
Canada 0.01 49
China 0.01 84
Austria 0.01 55
Japan 0.01 79
Singapore 0.01 56
Taiwan 0.01 76
Wide
Pak
Irel
180
Guat Mor
Pol
Kazak
Uru Tur Indo
Costa
Tight
SoAf Mex
Peru Rus
Isrl
Colo
100 Slvk
Brz
Pan
Slvn
Chin
Mala
Chil
56
0.03 0.1 0.3 1 3 10 30
Market-Implied PD (%)
Jama
Arg
Port Ukra
Hung
Viet
DomR Jord
Leba
ElSal
Sril
Icel
Spai Roma
Latv Thai
Ven
Gree
Sources: Citi Investment Research & Analysis and Markit Partners, LTD
28 January / February 2012

from bonds of major sovereigns are converted to their oneyear
equivalents and those spreads and spread volatilities
were input to Equation 4 along with an assumed LGD = 60
percent to produce market-implied PDs for a large number
of global sovereigns (i.e., those with at least one asset having
a three-month series of daily bond prices). 8 The resulting
PDs for August 1, 2011 appear in the table and graph
in Figure 6 along with the five-year CDS spreads (where
available) for those sovereigns. Not surprisingly, given the
turmoil in the peripheral European countries, Greece was
the riskiest sovereign tested, with a 39 percent implied
one-year PD, with Ireland and Portugal next riskiest, at 14
percent. Also, notice that PDs for Belarus, Venezuela and
Pakistan were all greater than 10 percent.
Sovereigns’ five-year CDS spreads are plotted versus
their log market-implied PDs in the inset plot of Figure
6. Notice first that the points are fit well by a straight line,
but that most points fall on either side of the line. If the
market-implied PDs are accurate, one could infer that
bonds from sovereigns plotting above the line are cheap,
Figure 7
5-Year Cumulative PD (%)
Five-Year Cumulative Default Rates By Agency Rating
On Logarithmic (Left) And Linear (Right) PD Axes
100 50%
10
1
0.1
AAA/Aaa
AA+/Aa1
AA/Aa2
AA-/Aa3
Log PD (Left Axis)
Linear PD (Right Axis)
A+/A1
A/A2
A-/A3
BBB+/Baa1
Sources: Moody’s Investors Service and Citi Investment Research & Analysis
BBB/Baa2
offering greater spread than average for their risk, whereas
points below the line are rich. It is important to note, when
BBB-/Baa3
BB+/Ba1
BB/Ba2
BB-/Ba3
B+/B1
B/B2
B-/B3
>=CCC/Caa1
40%
30%
20%
10%
0%
5-Year Cumulative PD (%)
Figure 8
Indicative Data On Ratings, Returns And Weights Of Citigroup’s WGBI Components
And Weightings From Proposed PD-Weighted Index Schemes
Trading
Symbol
S&P
Rating
1-Year
PD
(%)
12-Mo.
Return
($-
Hedged)
Original
Weight
US$
(MM)
3-Month Average PD-Weighted Indexes
Original
Daily Volume Standard
Multiplicative
Index
Weight Linear Log Linear Log
Australia AAA 0.01 5.00% 174,698 0.89% 0.94% 0.96% 0.90% 0.95%
Germany AAA 0.04 3.08% 1,279,912 6.53% 6.86% 6.61% 6.60% 6.18%
Canada AAA 0.01 5.27% 374,883 1.91% 2.01% 2.06% 1.94% 2.03%
Denmark AAA 0.01 3.54% 125,958 0.64% 0.68% 0.69% 0.65% 0.68%
France AAA 0.01 0.74% 1,349,695 6.88% 7.24% 7.40% 6.97% 7.31%
Finland AAA 0.01 2.16% 83,504 0.43% 0.45% 0.46% 0.43% 0.45%
Ireland BBB+ 8.14 3.68% 99,120 0.51% 0.35% 0.38% 0.31% 0.23%
Japan AA- 0.01 0.97% 6,383,673 32.55% 34.26% 35.02% 32.96% 34.56%
Spain AA 0.70 –0.85% 581,766 2.97% 3.03% 2.57% 2.90% 2.00%
Malaysia A 0.01 1.16% 66,093 0.34% 0.35% 0.36% 0.34% 0.36%
Netherlands AAA 0.01 2.81% 345,264 1.76% 1.85% 1.89% 1.78% 1.87%
Norway AAA 0.04 3.34% 40,006 0.20% 0.21% 0.20% 0.21% 0.19%
Belgium AA+ 0.11 –2.62% 353,079 1.80% 1.89% 1.72% 1.81% 1.52%
Poland A 0.20 1.82% 104,309 0.53% 0.56% 0.49% 0.53% 0.42%
Portugal BBB- 16.58 –20.90% 88,709 0.45% 0.15% 0.32% 0.08% 0.17%
Austria AAA 0.01 1.95% 236,500 1.21% 1.27% 1.30% 1.22% 1.28%
Italy A 1.76 –7.01% 1,221,374 6.23% 6.08% 5.11% 5.76% 3.68%
Singapore AAA 0.01 4.23% 53,326 0.27% 0.29% 0.29% 0.28% 0.28%
Sweden AAA 0.02 4.36% 91,633 0.47% 0.49% 0.49% 0.47% 0.47%
Switzerland AAA 0.02 4.71% 62,228 0.32% 0.33% 0.33% 0.32% 0.32%
United Kingdom AAA 0.01 9.33% 1,086,405 5.54% 5.83% 5.96% 5.61% 5.88%
Mexico A- 0.08 5.00% 115,515 0.59% 0.62% 0.57% 0.59% 0.51%
United States AA+ 0.01 5.20% 5,292,078 26.99% 28.40% 29.03% 27.32% 28.65%
Market Value ($-Billions) 19,610 20,424 20,439 20,587 21,268
Average PD 0.26% 0.20% 0.21% 0.18% 0.15%
Sources: Citi Investment Research & Analysis and Citigroup Index LLC
www.journalofindexes.com January / February 2012
29

Figure 9
Relative Change In Weight (%)
Percent Changes In Country Weights By PD-Weighting
Strategy For Countries Ranked By PD
10
0
-10
-20
-30
-40
-50
-60
-70
Portugal
Ireland
Italy
Spain
■ Linear PDs
■ Log PDs
Poland
Belgium
Mexico
Norway
Germany
Sources: Citi Investment Research & Analysis and Citigroup Index LLC
Switzerland
Sweden
comparing market-implied PDs for these sovereigns, that
this analysis assumes a recovery value of 40 percent for
all these credits, the same as that for the corporate credits
used to calibrate the model. 9 In fact, the model can easily
accommodate other recovery value assumptions; since λ t
is the same regardless of LGD, given a one-year spread,
spread volatility and assumed LGD ≠ 60 percent, one can
solve for p T
in Equation 4.
A PD-Weighted World Government Bond Index
A multitude of indexes for both equity and debt have
been proposed, with most major financial institutions
having their own suite of offerings. The World Government
Bond Index (WGBI), Citigroup’s flagship fixed-income
index, comprises investment-grade sovereign debt from
23 countries, denominated in 14 currencies. Inclusion in
the WGBI is dependent upon market size, credit quality
and investability. As is typical of many credit indexes,
the obligor and asset composition is adjusted monthly to
include new issues and eligible markets and to exclude
markets and issues no longer meeting index criteria. One
problem for index managers and investors is that monthly
index adjustments can introduce large discrete changes,
requiring costly buying and selling of securities in order to
rebalance. In the WGBI, this can be particularly severe if a
large sovereign issuer deteriorates and gets excluded from
the index (e.g., Greece in July 2010), or an improving, but
still risky, major sovereign issuer is added (e.g., Mexico
in October 2010). To minimize these discrete changes in
index portfolio composition, we introduce several potential
methods for constructing indexes weighted by default
probabilities and evaluate their relative merits. Although
we describe the method in the context of the WGBI, the
method is general enough to be applied to any index with
a credit-based criterion for inclusion.
As of October 1, 2011, the WGBI includes the 23 government
bond markets of Australia, Austria, Belgium,
Canada, Denmark, Finland, France, Germany, Ireland,
Italy, Japan, Malaysia, Mexico, the Netherlands, Norway,
Singapore
Australia
Canada
Denmark
France
Finland
Japan
Malaysia
Netherlands
Austria
United Kingdom
United States
Figure 10
■ United States 28.7%
■ Mexico 0.51%
■ United Kingdom 5.88%
■ Switzerland 0.32%
■ Sweden 0.47%
■ Singapore 0.28%
■ Italy 3.68%
■ Austria 1.28%
■ Portugal 0.17%
■ Poland 0.42%
■ Belgium 1.52%
■ Norway 0.19%
PD-Weighted WGBI
■ Australia 0.95%
■ Germany 6.18%
■ Canada 2.03%
■ Denmark 0.68%
■ France 7.3%
■ Finland 0.45%
■ Ireland 0.23%
■ Japan 34.5%
■ Spain 2.00%
Sources: Citi Investment Research & Analysis and Citigroup Index LLC
■ Malaysia 0.36%
■ Netherlands 1.87%
Poland, Portugal, Singapore, Spain, Sweden, Switzerland,
the United Kingdom and the United States. A detailed
description of the index, its inclusion criteria and other
major features can be found in Citigroup [2011]. Briefly, the
outstanding value of a market’s eligible issues must total at
least $50 billion, €20 billion or ¥2.5 trillion to be considered
eligible for inclusion. The WGBI is an investment-grade
index. Each obligor’s credit quality must be rated at least
A-/A3 by S&P or Moody’s to enter the index and at least
BBB-/Baa3 thereafter to remain in the index. Furthermore,
a market should limit barriers to entry by actively encouraging
foreign investor participation. Finally, only issues
with maturities longer than one year are included and the
index profile is updated on a monthly basis.
Because there is little precedent for PD-based credit
indexes, we tested two different schemes for weighting
market-implied PDs: algebraic and geometric. Although
a simple arithmetic weighting of PDs seems more natural,
there are reasons to suggest that a geometric weighting
scheme might be more appropriate. Consider, for example,
the five-year cumulative default rates by rating category
that appear in Figure 7. The blue bars, referenced
to the left axis, are the default rates scaled logarithmically,
whereas the connected red circles referenced to the
right axis are the same data on a linear scale. The figure
shows that the relationship between rating category and
default probability is better described by log PDs than
linear ones. The importance of that relationship for index
construction is that simple averages of PDs will not correspond
to the average rating category from the sample of
obligors from which they came. Rather, the data in Figure
7 imply that the average of countries’ log PDs will better
correspond to the PD of their average rating. 10 Thus,
we tested and compared the arithmetic and logarithmic
approaches focusing on the relative advantages of each.
The 23 countries in the WGBI appear in the left column
of Figure 8, followed by their Standard & Poor’s credit
ratings, one-year PDs, 12-month returns and U.S. dollarequivalent
values. The right columns of Figure 8 show percentage
weightings of the current index and four proposed
PD-based methodologies. Aside from the use of linear
30 January / February 2012

or logarithmic PDs, our weighting algorithm was similar
for each scheme. Within both linear and logarithmic
PD schemes, we examined the effects of two alternative
weighting adjustment procedures: a standard method and
a multiplicative one. Each is described in turn below.
The standard PD weighting algorithms begin by calculating
the averages of either PDs or logarithms of the
PDs. 11 That is, the linear PD average is calculated as:
where PD i
is the market-implied default probability for
the i th obligor and N is the number of obligors, which is
currently 23. For the logarithms of PDs, the average calculation
becomes:
where log indicates the logarithm of base10 and the other
terms are as defined earlier.
Having obtained the mean of the PDs and log PDs for
the countries in the WGBI, we calculate each country’s
distance from the given average as:
The distance measure is the same for linear and log PDs,
except that the log(PD i
) and μ log(PD)
are substituted for PD i
and μ PD
in Equation 7. Then, for either method, given the
distance measures, d i
for each country, the adjustment to
the dollar-weighted face amount is calculated as:
so that the adjusted face values for the PD-weighted
schemes are expressed as:
(9)
where M i
is the dollar-equivalent index weight of country i
in the index, with the superscripts O and PD indicating the
weight in the original index and the weight in the PD-weighted
indexes (either linear or logarithmic), respectively.
The resulting values of M i
PD
for the standard linear
and logarithmic weighting schemes appear in Figure 8
in the columns to the immediate right of that labeled
“Original Index Weight.” For comparison, the relative
change in each country’s percent contribution for the
linear and logarithmic PD strategies is plotted in Figure
9. The average PD of the 23 sovereigns in the WGBI is
1.27 percent, whereas the antilog of the average log 10
PDs
is 0.05 percent. 12 For both strategies, countries with PDs
lower than the average will be overweighted, and those
with PDs above average will be underweighted. Thus, as
shown by the red bars in Figure 9, since only Italy, Ireland
and Portugal have PDs above the average of 1.27 percent,
these are the only obligors that receive underweighting
(5)
(6)
(7)
(8)
in the linear strategy. However, given the smaller mean
of the log PDs, adjustments from the logarithmic strategy
are more balanced across obligors, such that Spain,
Poland, Belgium and Mexico are also underweighted.
Furthermore, because of the greater number of sovereigns
below the geometric (i.e., logarithmic) average, the
underweighting is less extreme than for the linear case.
Figure 10 shows what the actual country weightings of a
PD-weighted WGBI would be.
As mentioned above, we also examined linear and logarithmic
PDs with multiplicative weight adjustments. For
those adjustments, Equation 8 was modified to include a
multiplicative constant, k, such that:
(10)
Sovereign contributions determined using the linear
and logarithmic multiplicative adjustments appear in
the rightmost columns of Figure 8, respectively. 13 The
examples shown used multiplicative factors of k = 1.2 for
the linear PDs and k = 2.0 for the logarithmic ones. Figure
8 demonstrates that the multiplicative factors exaggerate
the adjustments, both up and down, of the weightings.
Currently, there is little reason to prefer a multiplicative
scheme over the simpler weighting formula in Equation
8. However, historical tests of how the various weighting
schemes affect index rebalancing and resulting riskadjusted
returns should provide information on this
issue. The historical PD estimates necessary to perform
those analyses are currently under way.
Summary
This report summarizes an approach derived by
Benzschawel and Lee [2011] for deriving PDs from obligors’
credit spreads and corresponding spread volatilities
calibrated to PDs obtained from credit models and
agency ratings. The method assumes that, on average,
credit spreads are linear functions of spread volatility
and that investors require the same level of spread
compensation per unit of spread volatility regardless
of its source. Evidence was presented to support those
assumptions, and market-implied PDs were generated
for sovereign credits to demonstrate the usefulness of
the model for tracking market perceptions of firms’
credit risk. These market-implied PDs can also be used
to construct credit schemes, based on indexes in which
obligor contributions are adjusted for default risk.
Examples of proposed adjustment schemes, based on
Citigroup’s World Government Bond Index, were given
to illustrate the advantages of PD-based weighting.
Market-implied PDs for global corporate and sovereign
credits are calculated daily and are available for analysis
in the Yield Book, Citigroup’s fixed-income analytics
platform.
continued on page 62
www.journalofindexes.com January / February 2012
31

Optimizing Fixed-Income
Index Funds
Trade-offs in portfolio construction and management
By Stephen Laipply and Christopher Woida
32 January / February 2012

The primary role of an index fund manager is to track
the fund’s designated benchmark. In practice, it is
difficult to obtain near-zero tracking error, unless
the manager holds every single security in the benchmark
and perfectly synchronizes portfolio rebalancing activity
with benchmark changes at minimal cost.
Holding every security in the benchmark can be challenging
in liquid markets, such as exchange-traded equities.
In less liquid markets, such as the over-the-counter
bond markets, it is virtually impossible. Because of the
discontinuous liquidity and wide bid/offer spreads often
present in the OTC fixed-income markets, holding more
securities may reduce tracking error in theory, but may also
ultimately result in much higher transaction costs.
Many fixed-income index fund managers seek to address
this problem by holding an optimized subset of available,
liquid securities. This strategy strives to balance market
risk and portfolio management costs using risk models
and optimization techniques to locate the intersection of
projected tracking error and projected transaction costs. In
short, the manager will attempt to drive projected tracking
error down to the point where any further decreases would
be outweighed by increases in projected transaction costs.
However, this conventional solution does not account
for the “short volatility” risk inherent in sampled portfolios.
Optimized portfolios rely heavily on assumptions about
asset volatilities, correlations and transaction costs. Rising
volatility often leads to a destabilization of correlations, an
increase in idiosyncratic risk and a widening in bid/offer
spreads. The combined effects (and their correlation with
each other) often result in higher-than-expected realized
tracking error in an environment of rising volatility.
In this paper, we test and quantify trade-offs among portfolio
solutions through historical market cycles to identify
a portfolio construction and management approach that
minimizes realized tracking error across a range of volatility
regimes and market conditions.
Portfolio Construction And Management
Fixed-income index portfolio managers strive to minimize
realized tracking error by targeting projected tracking
error (PTE) and transaction costs. PTE is typically defined as
the forecast standard deviation of the performance differential
between a portfolio and its benchmark. This projected
variation is based on a risk model and specific assumptions
about market parameters such as correlations and volatility.
PTE consists of common-factor risk (i.e., risk that is common
to all securities and therefore cannot be diversified away on
an absolute basis) and idiosyncratic (or, security-specific)
risk. While common-factor risk relative to a benchmark may
be eliminated by holding a relatively small sample of securities,
reducing idiosyncratic risk relative to a benchmark is
far more difficult and costly because a significantly larger
sample set of securities may be necessary. The larger the
sample size, the greater the proportion of less liquid securities,
which, in turn, increases transaction costs.
Traditional optimized portfolio construction strategies
rely heavily on the stability of correlations among portfolio
constituents (e.g., constructing a portfolio to minimize
tracking error based on a historical covariance matrix).
The challenge with this approach is that shifts in correlation
often invalidate the original optimized solution. As a
result, the portfolio manager is forced to choose between
incurring higher transaction costs in order to rebalance
to the new optimized solution or facing higher potential
tracking error by choosing not to rebalance. 1 Overreliance
on correlations—which the financial crisis proved is not
an advisable practice—can be mitigated by employing
rigorous stratified sampling techniques to construct index
tracking portfolios. Stratified sampling seeks to identify
and quantify index risk exposures that are not dependent
on correlation assumptions and can be matched through
judicious portfolio construction. A portfolio construction
approach based on stratified sampling techniques will
potentially be more stable and less vulnerable to sudden
shifts in asset correlations during market dislocations.
Shifts in correlation often occur in an environment of
rising volatility, which also may result in an increase in
idiosyncratic risk due to an increase in sampling error. If a
portfolio manager is holding a subset of securities relative
to the benchmark in an optimized portfolio (and is therefore
exposed to idiosyncratic risk because not all benchmark
holdings are represented), higher volatility will likely
result in higher sampling error. Conversely, at the limit,
zero volatility should result in zero sampling error.
The impact of higher volatility on tracking error was
pronounced in 2008 and 2009. As the financial crisis
unfolded, index optimization strategies incurred large
realized tracking error due to a destabilization in correlations,
an increase in sampling error/idiosyncratic
risk, and an increase in transaction costs. Figures 1 and
2 illustrate the volatility in transaction costs and tracking
error during the crisis.
Sampling error may only be mitigated by increasing
Figure 1
Liquidity Cost Score
Transaction Costs Observed During The Financial Crisis
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
Jan
2007
Jul
2007
Jan
2008
Jul
2008
Jan
2009
Jul
2009
Jan
2010
Jul
2010
US Credit 1-3Yr US Long Credit US Credit US Intermediate Credit
Source: Barclays Capital Liquidity Cost Scores
Note: The Liquidity Cost Score measures the cost (in basis points) of immediately
executing a round-trip transaction for a standard institutional trade using trader
bid/ask spread data, in order to gauge liquidity. It is computed as the difference
between bid and ask, multiplied by the spread duration of the bond.
www.journalofindexes.com January / February 2012
33

Figure 2
Tracking Error (bps)
Figure 3
Portfolio Operating Curve:
Annualized Transaction Costs As A Function Of Target PTE
Transaction Costs
0
Tracking Error Observed During The Financial Crisis
250
200
150
100
50
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Min 25 Percentile 50 Percentile 75 Percentile Max
Source: eVestment Alliance (Barclays Capital US Aggregate Bond Index)
Full replication
Projected Tracking Error
Liquidity-Biased Portfolio
the optimized sample size. Transaction costs, unfortunately,
are also a function of market volatility. As volatility
increases, transaction costs (as measured by bid/offer
spreads) tend to widen, as shown in Figure 1. The implication
is that, in a regime shift from a period of low volatility
to one of high volatility, it will become far more expensive
to respond ex post by increasing the optimized sample size.
In effect, this dynamic resembles a short-volatility position,
because both tracking error and transaction costs will
move directionally with market volatility.
Accordingly, it may be advantageous to hold a broader-than-prescribed
optimized sample in a low-volatility
environment (thereby incurring greater transaction costs
during rebalancing) to protect against regime shifts to a
higher-volatility environment later on. Incurring higher
transaction costs in a low-volatility base case is somewhat
similar to purchasing an option or an insurance policy. If
volatility remains low or falls, such a strategy may look
expensive relative to a strategy of not holding any protection.
However, if volatility rises and is accompanied by
correlation shifts and increases in idiosyncratic risk and
transaction costs, such a strategy would be vindicated.
To determine the effectiveness of this approach, we look
at historical market cycles. Before doing so, we first develop
a conceptual framework of the relationship between
projected tracking error and transaction costs.
Projected Tracking Error Vs. Transaction Costs
The conceptual relationship between sample size, transaction
costs and PTE, referred to here as the “portfolio
operating curve,” is depicted in Figure 3.
Transaction costs are plotted as a function of the PTE
targeted by the fund manager. In this conceptual representation,
it is assumed that marginal transaction costs are
unaffected by position trading size and portfolio flows; that
is, client flow crossing opportunities are not contemplated.
Note that, while a full replication of the benchmark eliminates
PTE, it also results in the highest level of transaction
costs, due to a higher percentage of less liquid securities
with wider bid/offer spreads. Accordingly, we should see
significant marginal savings in transaction costs as the
manager moves from full replication to an optimized solution
using more liquid securities. This is represented by the
portfolio operating curve in Figure 3, as we move from left
to right along the PTE axis.
Similar to an efficient frontier, each point on the portfolio
operating curve corresponds to a passive portfolio targeting
a specific PTE relative to its benchmark that is available
to the manager. Points above this curve correspond to
portfolios that are feasible but less efficient (i.e., portfolios
that may be constructed at a similar PTE but using less liquid
securities with higher transaction costs); portfolios that
lie below the curve are not feasible. A floor on transaction
cost savings occurs at the point at which the fund liquidates
only those securities that leave the benchmark and reinvests
proceeds based solely on a lowest-transaction-cost
criterion. We refer to this as the liquidity-biased portfolio.
This scenario, which incurs significant PTE, represents the
limit to which transaction costs can be reduced.
A key question is how portfolios along this curve
perform in various market environments. To find out,
we constructed and backtested different portfolio solutions
along the portfolio operating curve relative to the
appropriate market benchmarks and estimated the realized
tracking error associated with each. 2 Portfolios were
simulated and rebalanced monthly, with the objective
of minimizing a combination of PTE and transaction
costs over the backtested horizon. Figure 4 is a conceptual
illustration of transaction costs, PTE and realized
tracking error, based on actual relationships that were
observed in these backtests. Realized tracking error is a
function of the fund performance relative to the index
as well as transaction costs incurred through portfolio
rebalancing. For a given point on the portfolio operating
curve, the length of the arrow emanating from that
point (i.e., the distance between the point and the end of
the arrow) represents the largest single-month realized
tracking error observed over the historical backtested
horizon. This chart illustrates the relationship between
PTE and realized tracking error, and the potential pitfalls
associated with optimized solutions that emphasize
34 January / February 2012

Figure 4
Transaction Costs
0
Portfolio Operating Curve Vs. Largest
Single-Month Realized Tracking Error Observations
Full replication
Projected Tracking Error
Largest
single-month
realized
tracking error
Liquidity-Biased Portfolio
transaction-cost minimization at the expense of PTE.
The higher the target PTE, the greater the potential for
high realized tracking error during periods of elevated
volatility due to shifts in correlations, increased idiosyncratic
risk and increased transaction costs resulting from
managers attempting to mitigate increased tracking
error by expanding optimized portfolios.
Although portfolios with lower PTE appear to be more
expensive ex ante due to higher projected transaction
costs, they may ultimately prove to be less expensive
and more efficient ex post due to changes in market
conditions. Hence, it may be beneficial to ensure against
adverse market environments by holding a greater sample
of securities (and thereby incur higher baseline transaction
costs) than is typically prescribed by the conventional
optimization approach. By doing so, the portfolio
manager may ultimately spend less on future transaction
costs while experiencing a smaller divergence between
projected and realized tracking error in a higher-volatility
environment. We illustrate this concept through two
asset class examples: investment-grade credit and U.S.
Treasurys. These sectors were chosen both for their relevance
to fixed-income investors and because they highlight
specific trade-offs in portfolio construction.
Example 1: 1-3 Year Investment-Grade Credit
Consider a fund benchmarked to the Barclays Capital
U.S. 1-3 Year Credit Index. This benchmark typically
contains more than 600 securities and over 300 issuers.
Portfolio solutions were simulated over a backtested horizon
from 2007 to 2010. The inputs for the portfolio construction
process were the risk factors of the index (e.g., key
rate durations and spread durations). The common factor
covariance matrix and idiosyncratic risk estimates were
derived from time-weighted historical data. Because data
on bid/offer spreads in credit sectors is either not available
or not robust, transaction costs are modeled based on
Barclays Capital liquidity cost scores. Additional assumptions
for the simulations include:
• Full investment
• No minimum trade size
Figure 5
Mean transaction costs (bps/year)
Barclays Capital US 1–3 Year Credit Bond Index
Portfolio Operating Curve Vs. Largest Single-Month
Realized Tracking Error Observations
(2007–2010, Backtested Results)
45
Full Replication
40
35
30
25
20
15
10
5
Chosen Portfolio Solution
0 0 20 40 60 80 100 120 140
Mean projected tracking error (bps/year)
Source: BlackRock and Barclays Capital
Note: Mean PTE is expressed in bps per year, while largest single-month realized
tracking error is in absolute bps (not annualized).
• Monthly rebalancing
• Trading limited to rebalancing activity
(i.e., potential inflows are not incorporated)
• No out-of-benchmark holdings
• Transaction costs (in basis points) are constant
(i.e., no market impact due to trade size)
• Portfolio manager does not actively take advantage
of liquidity events
The resulting portfolio operating curve for this benchmark
is shown in Figure 5. First, note that the absolute level
of transaction costs in this credit sector is relatively high for
full replication (over 43 bps per year with 83 percent annualized
turnover). Due to the level and dispersion of transaction
costs for corporate securities (which represent 80 percent
of the market value of this index), there is a substantial
marginal savings in transaction costs for each increase in
target PTE, which is evidenced by the steep drop in the
portfolio operating curve with respect to increasing PTE.
Note that each increase in PTE corresponds to fewer
securities. Along this curve, the fund is still holding most of
the issuers in the benchmark but tends to overweight more
liquid issues. As an example, General Electric (the issuer)
had 40 different issues in the Barclays Capital U.S. Credit
Index, as of March 31, 2011. Five of these issues came to
market in the previous 12 months, with the remainder
more than 12 months old. Figure 6 shows the relationship
between bid/offer spreads and seasoning for GE issuance.
There is an explicit trade-off to this strategy in the form
of idiosyncratic risk. In periods of elevated market volatility,
the realized tracking error, driven by the idiosyncratic
risk between liquid on-the-run and less liquid off-the-run
issues (i.e., sampling error), can be quite high and can differ
significantly from PTE, as evidenced by Figure 5. This
exhibit depicts the realized tracking error for a given portfolio
solution along the transaction cost versus PTE operating
curve, and illustrates the risk of liquidity-biased solutions
in the presence of idiosyncratic risk and market volatility.
Accordingly, the portfolio manager would use the infor-
www.journalofindexes.com January / February 2012
35

Figure 6
Bid/Offer Spread (bps)
Figure 7
Cumulative Tracking Error (bps)
Bid/Offer Spreads Vs. Seasoning For General Electric
80
70
60
50
40
30
20
10
60
40
20
0
-20
-40
-60
-80
-100
-120
-140
0
0 10 20 30 40 50 60 70 80 90
Current age / Original term (%)
Sources: BlackRock and Barclays Capital
Barclays Capital US 1–3 Year Credit Bond Index: Selected
Portfolio Solution Vs. Full Replication Cumulative Realized
Tracking Error (Backtested Results)
-160
Mar
2007
Aug
2007
Jan
2008
Jun
2008
Nov
2008
Apr
2009
Sep
2009
Mar
2010
Full Replication Portfolio Cumulative Realized Tracking Error
Selected Portfolio Cumulative Transaction Costs
Selected Portfolio Cumulative Realized Tracking Error
Selected Portfolio Cumulative Performance Differential
Source: BlackRock and Barclays Capital
mation in this chart to trade off PTE, transaction cost
estimates and historical realized tracking error in order to
determine an optimal portfolio solution across different
volatility regimes. Assume, for example, that the manager
selected a solution with 9 bps per year average PTE and 24
bps per year of estimated transaction costs, corresponding
to 71 percent annualized turnover (the “chosen portfolio
solution” in Figure 5). While solutions with much lower
transaction costs exist with favorable trade-offs in PTE, the
worst-case historical realized tracking errors for such solutions
are far greater than PTE would suggest, and would
easily overwhelm any perceived transaction cost savings.
Figure 7 illustrates the backtested results of the chosen
solution. The horizontal axis represents time and the vertical
axis represents cumulative realized tracking error.
Note how the blue line corresponding to the fully replicated
portfolio steps down each month due to incurred
transaction costs related to turnover in the benchmark
(by definition, there is no performance differential with
the fully replicated portfolio—the only source of tracking
error for such a portfolio is transaction costs). The red line
corresponding to realized transaction costs in the sampled
portfolio solution is above the full replication line. This is
because the sampled portfolio solution trades off lower
transaction costs for higher PTE, which decreases annualized
turnover (from 83 percent to 71 percent).
In addition to transaction costs, each passing month
results in a cumulative performance differential relative
to the benchmark (the black line) for the chosen
sampled solution. The green line—cumulative realized
tracking error—represents the sum of the cumulative
realized transaction costs and cumulative realized
performance differential between the portfolio and the
benchmark. Note that the chosen portfolio solution
was superior to a full replication because the portfolio
manager was able to successfully trade off tracking
error with transaction costs.
This example demonstrates the importance of incorporating
the potential impact of different volatility
regimes in less liquid asset classes. The level of idiosyncratic
risk increases dramatically as we move from
relatively homogenous sectors, such as U.S. Treasurys,
to more heterogeneous sectors, such as investmentgrade
and high-yield credit. Credit sectors in particular
are prone to periods of dislocation, resulting in sudden
shifts in correlations, sharply increasing transaction
costs, and reduced liquidity in volatile market conditions.
A liquidity-biased portfolio solution is likely to
experience far more realized tracking error in an environment
of elevated volatility; the portfolio manager, in
an effort to lower idiosyncratic risk, may find it too costly
to increase the sample size of securities after market
conditions have deteriorated.
With this understanding of the trade-off between PTE,
estimated transaction costs and historical realized tracking
error, a set of portfolio management guidelines could
be developed. Using the backtested results, corresponding
portfolio risk attributes and performance statistics could
be compiled and tolerances for aggregated risk could be
generated using dimensions more intuitive to the portfolio
manager (e.g., key rate durations, spread durations and
concentration measures). Because the tolerances for these
risk dimensions would be based on observations compiled
over the backtested horizon covering a variety of market
cycles, they could serve as a useful guide in managing
portfolio risk over a range of volatility regimes and market
conditions. These guidelines could be used as heuristics by
the portfolio manager for monitoring risk on a daily basis.
Example 2: 7-10 Year U.S. Treasury Fund
Now consider a fund benchmarked to a U.S. Treasury
index, the Barclays Capital U.S. Treasury 7-10 Year Bond
Index. This benchmark typically contains approximately
20 securities (notes recently issued with 10 years to maturity
and older securities originally issued at longer maturities).
Portfolio solutions were simulated over a backtested
horizon from 2005 to 2010. The inputs for the portfolio
construction process were the common risk factors of the
36 January / February 2012

Figure 8
Mean Transaction Costs (bps/year)
Barclays Capital US Treasury 7–10 Year Bond Index
Portfolio Operating Curve Vs. Largest Single-Month
Realized Tracking Error Observations
(2005–2010, Backtested Results)
2.05
2 Full Replication
1.95
1.9
1.85
1.8
Portfolio A
1.75
Portfolio B
1.7
0 0.5 1 1.5 2 2.5 3 3.5
Mean projected tracking error (bps/year)
Sources: BlackRock and Barclays Capital
Note: Mean PTE is expressed in bps per year, while largest single-month realized
tracking error is in absolute bps (not annualized).
Figure 9
Barclays Capital US Treasury 7–10 Year Bond Index:
Portfolio Solution Cumulative Realized Total Tracking Error
Vs. Full Replication (Backtested Results)
Cumulative Tracking Error (bps)
2
0
-2
-4
-6
-8
-10
-12
Feb
2005
Sep
2005
Apr
2006
Nov
2006
Source: BlackRock and Barclays Capital
Jul
2007
Jan
2008
Aug
2008
Mar
2009
Oct
2009
May
2010
risk was defined as the volatility between on-the-run and
off-the-run yields and the fitted U.S. Treasury spot curve;
security exposure was defined by key rate durations. The
common factor covariance matrix and estimates of idiosyncratic
risk were derived from time-weighted historical
data. Transaction costs were determined by bid/offer
spreads for both on- and off-the-run securities. To capture
the impact from the financial crisis, transaction cost estimates
were doubled from July 2007 to September 2009;
in some instances, actual transaction costs for this period
were even higher. All other assumptions are the same as
those noted in the preceding example.
Results of the backtest are illustrated in Figure 8, which
depicts the portfolio operating curve for this Treasury index.
4
From a qualitative perspective, the results are mostly as
expected. Full replication has the highest transaction costs:
almost 2 bps per year with 48 percent annualized turnover.
There is a steep initial savings in transaction costs, as the
fund takes on risk by employing more sampled portfolios
(moving from left to right). Similar to Figure 5, the largest
single-month observations of realized tracking error for a
given portfolio also increase with the degree of sampling.
Unlike the preceding investment-grade credit example,
however, the curve eventually turns upward, indicating
that, despite the fund’s being managed at a higher level
of PTE, it would actually have cost more to stay within this
PTE range over the entire backtested horizon than it would
have to run the fund at a lower overall PTE target.
This effect is due to the increase in market volatility during
the financial crisis, which affected the management of this
fund in two specific ways. First, spreads of on- and off-the-run
U.S. Treasury yields relative to the fitted spot U.S. Treasury
curve dislocated and became much more volatile. This can
be thought of as an increase in idiosyncratic risk. Second,
bid/offer spreads across all U.S. Treasury securities widened,
driving transaction costs higher. To simply remain within the
target PTE during this period of elevated volatility, the portfolio
manager would have had to increase the sample size of
the portfolio to offset the increase in idiosyncratic risk. The
subsequent portfolio rebalance would have resulted in much
higher transaction costs than originally anticipated. These
transaction costs were higher due to both the nature of the
securities themselves (i.e., less liquid, off-the-run securities)
and higher volatility (resulting in wider bid/offer spreads for
these securities) observed over the backtested period. The
higher the initial PTE of the sample portfolio, the higher the
ultimate realized transaction costs and realized tracking error
observed over the backtested period.
The manager in this scenario was faced with a challenging
trade-off: By running more highly sampled portfolios,
they could either incur high realized tracking error or
attempt to reduce this realized tracking error by incurring
higher-than-anticipated transaction costs through rebalancing
to a broader portfolio. Either way, a better solution
would have been to incur higher baseline transaction costs
by holding a broader initial sample of securities and, therefore,
ultimately incur lower realized tracking error and
rebalancing costs (i.e., less rebalancing would be necessary
with a broader initial portfolio).
As an example, operating the fund at a PTE target of 1.1 bps
per year (Portfolio A in Figure 8) would indeed have resulted
in lower transaction costs in 2005 and 2006. However, these
early initial transaction cost savings would have ultimately
been eclipsed by higher realized transaction costs in 2007 and
2008, as the portfolio manager would have had to expand the
portfolio sample into less liquid securities in order to maintain
the original PTE target. Accordingly, from a realized tracking
error perspective, it would actually have been more efficient,
on average, to operate the fund closer to the benchmark at a
higher rate of turnover and a higher level of ex ante transaction
costs than to take on more PTE in an attempt to decrease
continued on page 64
www.journalofindexes.com January / February 2012
37

Another Look
Rethinking The Barclays Aggregate
As An Investment
The popular benchmark comes with some drawbacks
By Rich Wiggins
After a two-year uninterrupted trend of net inflows,
the amount of money flowing into bond funds has
exceeded the cash that went into stock funds during
the Internet bubble. Inflows slowed down a bit after Pimco
bond guru Bill Gross publicly turned uber-negative on U.S.
Treasurys—but not before investors poured more than $500
billion into bond funds, fueling a rally that eventually drove
two-year Treasury yields to an all-time low of less than 0.5 percent.
To put that into perspective, total bond net inflow for the
previous 10 years combined was only $423 billion. According to
the Investment Company Institute, while equity mutual funds
saw outflows of more than $63 billion, fixed-income funds saw
more than $87 billion in inflows during the first nine months of
2011. Much of this inflow was invested in a single index—the
Barclays Capital U.S. Aggregate Bond Index.
The bond market is more than twice as big as the stock
market, but there’s never been the sentiment that bonds
should be given the same attention to implementation as
stocks. Inexplicably, investors don’t show the same level
of consideration and differentiation to the bond portion of
their portfolio as they do their equity stakes. Fixed-income
portfolio management is very much an index-relative game:
Incremental sector, quality and duration bets are all made
with an eye to how the index is constructed. Very rarely is
a benchmark-unconstrained fixed-income manager seen.
Relative to equities, there’s a disproportionate tilt toward
completely passive investing in fixed income, and the Barclays
Capital U.S. Aggregate Bond Index is widely considered the
best benchmark for efficient asset allocation.
Origins
Charles Dow may have created his Dow Jones
Transportation Average way back in 1896, but total return
bond indexes weren’t developed until 1973. The great
portfolio innovations of the 1950s and 1960s that addressed
asset allocation could not be applied without a measure of
bond performance, so Art Lipson at Kuhn, Loeb created
a bond index to close this measurement void. Salomon
Brothers followed with a similar index two weeks later but
it never really caught on. By the time Lehman Brothers purchased
Kuhn, Loeb at the end of 1977, Art’s original index
had already become the bond benchmark, and history has
solidified this claim.
Known as the Barclays Capital U.S. Aggregate Bond
Index since Barclays Capital took over the index business
of the now-defunct Lehman Brothers, the “Barclays Agg”
comprises about $15 trillion worth of bonds and is designed
to include the whole landscape of domestic, investmentgrade,
fixed-income securities traded in the United States.
Employed in both active and passive approaches, this
“sacred cow” index is a flagship benchmark that plays an
integral part in the fixed-income world; and the stampede
of money flows into fixed income in recent years has made it
even more so. There are over 110 generic bond indexes but
practically everybody uses this one index.
Debt indexes have become not only measurement tools,
but financial products in themselves—just take a look at the
still-young ETF industry. The iShares Barclays Aggregate
Bond ETF (NYSE Arca: AGG) is the world’s largest bond
ETF, with assets of $13.8 billion, and considered by many
investors to be a “one-stop shop” for fixed-income exposure.
However, there are three other U.S.-listed ETFs tied to the
Barclays Agg; with AGG included, their total assets amount
to $27 billion as of the end of October 2011. That’s more than
16 percent of the total assets invested in U.S.-listed domestic
fixed-income ETFs—115 funds in all.
38 January / February 2012

Investors building a traditional 60/40 portfolio may own as
many as 10 different funds to cover their equity exposure, but
often have been using the Barclays Aggregate as a placeholder
for years—utilizing the fund as the sole component of the fixedincome
portion of their portfolio (and often not even considering
the international bond market, but that’s another topic entirely).
This index—which is basically the fixed-income analog
for the Russell 3000—is more than just a key touchstone in
the bond market. It underpins the strategic and tactical asset
allocation of billions upon billions of dollars, but America’s
unimaginably massive budget deficits combined with the
unprecedented government intervention in U.S. financial
markets in 2008 have meaningfully changed the nature of an
investment in this benchmark. Most investors are well aware
of the impact such astronomical borrowing has on the country’s
budget, but few have considered the impact these programs
have had on this fixed-income benchmark—an impact
that highlights one of the index’s key problems.
Encroaching Government
Fannie Mae and Freddie Mac were put into conservatorship
in September 2008, and ever since then have been wards
of the government. The placing of Fannie Mae and Freddie
Mac in government conservatorship effectively turned their
debt into government debt and immediately transformed
the U.S. Aggregate into predominantly a “government bond
index.” The index was roughly 35 percent government-affiliated
bonds at the end of 2007, as shown in Figure 1 (Treasury
debt + government-related debt). With that mortgage debt
added in at the end of 2010, that number was closer to 80
percent. 1 MBS debt, after all, represents the vast majority of
the securitized debt in Figure 1. And with each round of new
Treasury auctions, the Barclays Aggregate begins to more
closely resemble a dedicated government bond fund.
Investors often have a “my benchmark thinks for me”
attitude but, historically, it’s never been a good sign when
indexes become lopsided. In 1990, Japan dominated the
EAFE with approximately a 60 percent weight just in time
for a long bear market that would bring its weight down to
the vicinity of 20 percent. However, there’s complacency
around the Barclays Aggregate because so little “adverse history”
is associated with it. Unlike equity indexes, which have
experienced severe periodic crashes, the Barclays Aggregate
has posted only two negative calendar returns—in 1994 and
1999—since 1977, and those were relatively tame, at -2.92
percent and -0.82 percent, respectively. And yet if yields move
from 3 to 7 percent and corporate spreads widen to 350 bps
from 144 bps, the Barclays Aggregate index—which has an
average maturity of about 7 years—can be expected to correct
between 25-40 percent. This would be a catastrophic shock to
investors who bulked up on bonds thinking they were allocating
out of “risky” stocks and into “safe” bonds.
Not only are there a lot of agency mortgage-backed
securities in the index, but that market has been propped
up by massive artificial demand. Between November 2008
and March 2010, the Fed purchased $1.25 trillion of MBS,
representing approximately 25 percent of the outstanding
market. The implications of the end of the government’s
unprecedented support for mortgage-backed securities
remain uncertain. To an investment community concerned
about tail risk, it seems clear that it will have dramatic consequences
in the opposite direction. As markets normalize
and the government ceases to be the dominant source of
demand in the market, it leaves little upside potential but
tremendous downside risk. In November 2011, MBS represented
approximately 32 percent of the Barclays Agg.
Treasury, agency and agency mortgage-backed bonds,
which are supported by the U.S. government, have very low,
if any, credit risk. In fact, they were the holy grail of asset
classes during the financial meltdown of 2008 and again
during the European debt crisis last year. But that’s changing
as the Fed’s balance sheet has been growing every day;
so much so that on April 18, Standard & Poor’s dared to tell
Figure 1
Source: Barclays Capital
Barclays Capital U.S. Aggregate
Annual Sector Breakdown (%)
Treasury
Gov
Related
Corporate
Securitized
12/31/1990 45.50 10.11 15.11 29.27
12/31/1991 45.01 9.29 14.48 31.21
12/31/1992 45.26 9.17 14.52 31.04
12/31/1993 46.93 8.75 14.72 29.60
12/30/1994 47.01 8.89 13.98 30.13
12/29/1995 46.03 8.94 15.30 29.74
12/31/1996 45.06 8.66 15.68 30.61
12/31/1997 42.91 8.61 17.31 31.17
12/31/1998 37.93 10.84 19.39 31.84
12/31/1999 32.50 11.74 18.83 36.93
12/29/2000 26.77 14.08 20.58 38.57
12/31/2001 22.02 15.63 23.12 39.22
12/31/2002 21.86 16.91 22.27 38.96
12/31/2003 22.16 15.77 22.15 39.92
12/31/2004 24.68 15.16 20.65 39.51
12/30/2005 25.24 15.04 19.52 40.20
12/29/2006 24.68 14.72 19.40 41.20
12/31/2007 22.35 12.98 19.57 45.10
12/31/2008 25.07 13.54 17.67 43.72
12/31/2009 27.65 13.21 18.82 40.33
12/31/2010 33.75 11.97 18.79 35.49
the truth and lowered its outlook on U.S. debt to “negative”
from the much-more-appealing “stable” designation. The
next financial crisis will in all likelihood be entirely different,
so what was safe last time won’t be safe next time. If
tomorrow’s sovereign troubles include the United States,
then Treasurys could be one of the worst-performing assets.
That explains why a U.S. corporate bond traded in 2010 at a
lower yield level than a U.S. Treasury of the same maturity. 2
Investors see more risk in owning U.S. government debt
than owning debt of a company with a fortress balance sheet
like Berkshire Hathaway—prompting Morningstar to note
that, “The markets know no patriotism.”
It’s been said that more money has been lost reaching
for yield than has been lost in the stock market—and that
may be true—but reducing the overweight position in U.S.
Treasurys and mortgage bonds and adding more credit risk
continued on page 63
www.journalofindexes.com January / February 2012
39

In Perspective
Bond Indexing Ripe
For A Renaissance?
As demand from investors for fixedincome
investments grows, so too do
opportunities for index providers
By David Krein and John Prestbo
David Krein
John Prestbo
A
price is a price is a price. At least it is for a stock, but
not necessarily for a bond.
Whether you subscribe to an expensive Bloomberg
terminal or have bookmarked your favorite stocks on a
cellphone app, you can quickly pull up this afternoon’s
closing price for AAPL or GE. Historical price data
that stretches back years or even decades also can be
displayed in an instant.
This is possible because equity price data are conveniently
centralized in one or two primary exchanges per
market, making thousands of tickers available to all interested
parties. The information flows at warp speed to all
ends of the investment spectrum.
Clearly, the availability of current and historical data is
important, but its breadth, depth and validity are also critical
factors. After all, equity price data are official, objective
and universal. Equity price data are democratized.
Investors are empowered to research a market, develop an
investment thesis, tailor the approach, backtest the strategy
and monitor the positions going forward. These same characteristics
largely explain why equity indexes represent the bulk
of what index providers offer. Not only is indexing more established
in the equity world, but the providers can simply spread
the data feed overhead across more and more stock indexes.
It is a completely different story in the bond world,
where there are no exchanges, pricing transparency or
objectivity, or meaningful data availability.
Long ago, bond price data existed only in the large
investment banks that had the capital and technology to
make markets in such instruments. Even then, each bank
assembled prices expressly for securities in markets that it
participated in regularly. Further, all trading was done over
the counter by traders who knew each other and with institutions
that maintained a regular book of business.
Bond investors sometimes had to call two, three or
more dealers to find someone with the right mix of inventory
and interest to get a price. Even then, it was only one
price and nearly impossible to determine whether it was
competitive. If you had an obligation to get three prices
before transacting, well, that was a lot of dialing, and prices
would often fluctuate in the interim.
This approach applied for all but the most liquid bond
markets (and, more specifically, the most liquid bonds
within the most liquid bond markets). Focused on munis?
Interested in automotive corporates? Looking for highquality
sovereigns? There are dozens of markets, and each
had dozens, hundreds or even thousands of issuers. The
total number of bonds was (and still is) orders of magnitude
greater than the number of stocks.
From an indexing perspective, this has long been an
extremely difficult hill to climb. Simply navigating the logistics
of collecting and organizing the necessary data from so many
parties—and for so many instruments—was a Herculean task.
As a result, indexed offerings in the fixed-income markets
remained far behind those in the equity markets.
Moreover, rather than make the current and transaction-based
data available for lower-revenue indexing
activity, the banks preferred to keep it private and leverage
it for their own higher-revenue bond-trading activity. (One
notable exception here is Lehman Brothers, which used
its namesake indexing business to bolster its bond-trading
platform. Lehman, of course, has since been absorbed into
other large financial companies, with the bulk of its operations
and indexing business now part of Barclays Capital.)
Over time, banks got smarter and faster in pricing a
wider array of bonds and aggregating the historical data. In
40
January / February 2012

addition, interdealer bond brokers and data vendors have
expanded their presence in the real-time bond-price business,
helping to disintermediate the banks.
As a result, the times, they are a-changin’ for the indexers.
Bond-price data is now far easier to obtain. Clean, timely and
accurate data can be had from multiple independent sources,
even for narrowly defined markets with less investor interest.
The infrastructure and technology required to build and
maintain bond indexes has also improved dramatically, and
can be readily matched with other metrics such as fundamentals
or liquidity. Offerings can now move beyond the plain
vanilla (U.S. investment-grade corporate bonds, for example)
toward indexes that take real-world investment demands into
account, such as nondollar and emerging markets bonds.
Further, investor interest in these indexes tied to products—
such as ETFs/ETNs, structured products and institutional
funds—ultimately strengthens the providers’ economics.
After all, bond market investors—many of whom remain
overweight in active strategies—want the same benefits
that typically accrue to equity market index investors:
improved transparency, predictable factor exposures,
higher liquidity, longer track records with backtesting,
and, not to be forgotten, lower costs. Put together, these
play very favorably to bond investors’ advantage. All they
need are indexed investment choices.
So although indexers do not have the same legacy in the
bond market that they have in the equity market, do the
indexers currently have anything to contribute to the fixedincome
marketplace at this stage? An examination of the
data suggests that the answer is an unequivocal “yes.”
According to the 2011 Investment Company Fact Book,
365 U.S. index mutual funds as of year-end 2010 managed
total net assets of $1 trillion. Of that $1 trillion, 81 percent
($810 billion) was invested in domestic and international
stock indexes. (This $810 billion is a 14.5 percent share of
all equity mutual fund assets, up from 5.2 percent in 1996.)
The remaining 19 percent ($190 billion) was invested in
bond indexes or hybrid funds, yet attracted 40 percent of
the new money that flowed to index funds that year.
There are several compelling observations to be made
in analyzing these statistics. First, while the ratio of equity
mutual fund assets to bond mutual fund assets in the
U.S. is nearly 2-to-1, the ratio of stock vs. bond index
fund assets is nowhere close to that proportion—clearly
suggesting that there is plenty of room to grow the bond
index fund share of total index funds.
Second, the assets in bond index funds as a percentage
of the total assets in bond funds are far below that on the
equity side of the fence. Thus, the share of indexed assets
within the bond fund category can expand as well.
Third, the portion of indexed assets within the mutual
fund space continues to climb, implying an opportunity for
continued overall growth.
Fourth, in collecting 40 percent of new money for index
funds, the demand for bond index funds is clearly quite
high. Of course, the demand for fixed-income investments
overall has risen, as investors have sought less-risky assets in
response to the market turmoil of the last few years, and this
has apparently been a boon to bond index fund inflows.
(Similar cases can be made with regard to ETFs and
institutional funds. Although the data are not as skewed for
mutual funds, they still reinforce the notion that bond index
market share has plenty of room to grow across the board.)
With demand up and better data increasingly available,
the time is ripe for index providers to step up their game
in the fixed-income space. They are faced with an unprecedented
opportunity in a vast and largely unexplored asset
class. The fixed-income index space is nowhere near as
developed as the equities index space, and all the pieces
are currently in place for that to change dramatically.
The process has already begun, but you ain’t seen
nothing yet.
Why subscribe to the
The Journal of Indexes is the premier source for financial index research, news and
data. Written by and for industry experts and financial practitioners, it is the book of
record for the index industry. Browse content online at www.indexuniverse.com/JOI
TO SUBSCRIBE, VISIT:
www.indexuniverse.com/JOI/subscriptions
Redefining Credit Risk
William Mast
Credit Derivatives Indexes
Gavan Nolan and Tobias Sproehnle
A Fixed-Income Roundtable
Ken Volpert, Jason Hsu, Waqas Samad, Larry Swedroe and more
The Impact of Bond Fund Flows
David Blanchett
Plus David Blitzer on bubbles, Jeremy Schwartz on dividends and buybacks, Francis Gupta on country
classifications and a biography on Bogle
www.journalofindexes.com January / February 2012 41

Talking Indexes
Bespoke Investing
When the lines between asset classes
begin to blur
By David Blitzer
In 20th-century investing, stocks and bonds were different
and reasonably simple. Stocks were all about
capital gains, risks, dividends (sometimes) and opinions,
while bonds were about safety, income, ratings and a
little bit of mathematics. Many of these notions have been
altered since the appearance of ETFs and structured products
in recent years, giving us income-oriented products
based on stocks and risk-focused instruments built from
bonds. To some extent and for some investors, too many
choices mean confusion, not opportunity.
The good old simple days were easy to explain. Stocks
are a share in ownership and a claim on uncertain
profits—uncertain because everything else must be paid
before profits are earned. On the positive side, profits
may have virtually unlimited upside. That upside can
inspire wild dreams, and those dreams may turn into
stratospheric price gains even if the profits lie far in the
future. Of course, sometimes the rush to the stratosphere
becomes a bubble. Dividends encourage shareholders to
stay the course when few believe that profits will surge
and sustain share prices. Evaluating and choosing stocks
is as much a beauty contest as an analytical challenge.
Bonds represent a senior claim on revenues among all
those claims that come before profits. Bonds offer a promise
of both income and eventual return of the principal
investment with the solemnity of the promise being independently
rated. Some differences among bonds depend
on the mathematics of bond pricing: Bonds with unusually
large coupons or short lives respond very differently to
interest rate changes than seemingly similar bonds with low
coupons or long lives. Bonds can also differ by the ability of
the bond issuer to pay interest and principal as promised.
Compounding all these bond/stock differences were the
market structures where almost all stocks are traded almost
every day, so their prices reflected competing visions of the
future, while few bonds trade on any given day, and their
prices are often someone else’s best estimate.
ETFs—especially what one might call “second generation”
ETFs—have changed all this. First-generation ETFs,
such as those tied to the S&P 500, track a market or a
clearly defined market segment and, through market-cap
weighting, take on the investment and statistical properties
of the market or market segment. Their lure is the
efficient means to participate in the market—an investor
looking to invest in large-cap stocks got large-caps stocks.
On the bond side, there were some bits of subtlety since
an index of several thousand bonds might be tracked with
a portfolio of a few hundred issues, but the link to the
market’s behavior was still very strong.
Second-generation ETFs, however, are bluring the division
between stocks and bonds.Take low-volatility equities
as an example. The extreme gyrations experienced in
the equity markets in recent years worry many investors.
Turning to bonds to escape volatility would mean trading
hopes of capital gains in exchange for low coupon income.
An “engineered” alternative would be a low-volatility index
and associated ETF offering stocks selected and weighted
for reduced anxiety or churning. The methodology might
choose the 100 stocks of the S&P 500 that are the least
volatile as shown by their daily returns over the last year,
and then weight the index by the reciprocal of their volatility.
Instead of tracking the large-cap U.S. segment of
the stock market and all its statistical pluses, minuses and
quirks, this index extracts a key previously hidden aspect—
low-volatility equity behavior. The hopes for profits are
retained and combined with the stability that was previ-
42
January / February 2012

ing interest rates. While market risk can be adjusted by
selecting differing combinations of coupons and maturities,
default risk was pretty much fixed once the issuer was
chosen. The development of credit default swaps means
that default risk can easily be added or subtracted from an
underlying portfolio, or even added or subtracted without
the need of an underlying portfolio.
These efforts—and the examples here barely scratch
the surface in variety, complexity or creativity—give us an
“age of bespoke investing” where the properties of an ETF
Bonds offer income from periodic interest payments. ETFs built from
dividend-paying stocks and weighted by dividends or dividend yield
include a “promise” of income. One can create a bond-behaving ETF
with targeted income and stability, built solely from stocks.
ously regarded as the province of bonds.
One can take this process much further. Bonds offer
income from periodic interest payments. ETFs built from
dividend-paying stocks and weighted by dividends or dividend
yield include a “promise” of income. One can create
a bond-behaving ETF with targeted income and stability,
built solely from stocks. The mirror image is also possible—
building stocks from bonds. Recall that the origin of the
Black-Scholes option pricing model was a hedge equation
that replicated a stock as an option on its price and a dividend
stream based on a fixed-income instrument.
Index-linked structured products that promise a
proportion of an equity index return, often combined
with some downside protection, are built on the same
approach. An older example is an index-linked certificate
of deposit that pays either 65 percent of the S&P 500’s
total returns or 1 percent, whichever is greater. Other customizations
are possible with bonds. Traditionally, bond
risk was divided into credit risk (i.e., whether or not the
issuer would default) and market risk related to chang-
are not limited to the stocks or bonds that happen to be
in the market at the moment. Before the rise of bespoke
investing, the characteristics of an investment were determined
by the issuer’s needs; now it is increasingly possible
to tailor the characteristics to the investor’s desires.
However, there is one crucial aspect of any investment
that the age of bespoke investing cannot alter: Markets
are fickle, and sometimes the best-laid plans collapse.
There still are no free lunches even though the modern
menu now offers special diets and more variety.
Announcing the new
ETF Education Center
Reading
Qualifies for 1 hour of CFP CE Credit
Visit the ETF Education Center at:
www.indexuniverse.com/education
www.journalofindexes.com January / February 2012 43

Perspectives On Fixed-Income
Index Design
Comparing market-value-weighted benchmarks
with other bond index design alternatives
By Brian Upbin
44
January / February 2012

As an index provider, Barclays Capital occasionally
hears commentary from some observers that the
only fixed-income indexes investors have at their
disposal are market-value-weighted indexes and that they
are flawed by design. While it is true that market-valueweighted
bond indexes such as the Barclays Capital US
Aggregate Index are the most widely used benchmarks
for bond portfolios, this characterization of the fixedincome
index landscape is overly sensationalized, given
the breadth of offerings and variety of index designs
actively in use by investors looking for the most appropriate
indexes for their respective portfolios.
Philosophically, Barclays Capital recognizes that no
single index design is universal and, therefore, offers a
variety of solutions for investors. We are agnostic with
regard to the index choices made by investors; our various
methodologies are simply different ways of measuring
the same asset class in a rules-based, objective,
transparent and usable manner. In other words, “to each
(investor) his/her own (benchmark).”
The selection of the “right” fixed-income index consistently
boils down to selecting what is most appropriate
and usable for a specific portfolio objective (for dedicated
fixed-income portfolios and as part of an overall asset
allocation mix). In this article, we will outline a generalized
framework for investors to consider when selecting
an appropriate fixed-income benchmark; touch upon
the reasons why market-value-weighted indexes are so
widely used by investors; and discuss common benchmarking
alternatives for those who have portfolio objectives
that may require an index design option other than
a standard market-value-weighted index.
Framework For Bond Index Design And Selection
Common Uses for Fixed-Income Indexes
As an index provider, we come across three common
uses for fixed-income indexes, each of which may influence
preferences in index design.
• The most common use is as a baseline performance
target or benchmark for active or passive bond portfolios.
Some may think of benchmarks solely in the context
of performance analysis, but usage of benchmark index
data occurs at many stages of the portfolio management
process including asset allocation, security selection, and
ex-ante portfolio risk analysis.
• A second use for fixed-income indexes is as informational
measures of market performance and risk
characteristics; this information is used by a variety
of functions within a firm in the development, backtesting,
evaluation and implementation of investment
strategies and market analysis.
• Finally, fixed-income indexes are also used as a reference
for passive investment strategies and index-linked
products such as ETFs, ETNs, structured notes, etc.
Defining Index Objectives
Most standard market-value-weighted indexes are
suitable for any of these three use cases, but a particular
investor may have a specific portfolio objective that would
require a different index design. If choosing a bespoke or
less widely used benchmark, an index user will implicitly
address and answer a common set of questions that
would allow them to select the most appropriate index.
1) What is the fixed-income index trying to measure?
The initial question investors must address in their
benchmark selection process is what universe of securities
their chosen index should be measuring. Some
investors feel that an index should measure the full universe
of securities in their investment choice set, while
others feel a narrower universe of securities is more
appropriate for a specific portfolio objective. In many
cases, this question will be explicitly referenced in an
investment policy statement.
Those arguing for broad inclusion generally feel that
not investing in a particular asset class is a conscious
active management decision that is an attributable
source of portfolio returns and performance. Those
arguing against broad inclusion often cite that the rationale
that just because they can invest in an asset class
does not necessarily mean they should, and that excluded
asset classes or sectors should remain a tactical view
rather than an ongoing strategic decision.
Preferences on the scope of a chosen benchmark
are also a function of the flexibility an investor has to
invest in out-of-index securities and whether those
investments are tactical or strategic in nature. Investors
with more discretion may use out-of-index bets as a
source of additional risk-taking and the main source of
portfolio alpha, and may, therefore, prefer a narrower
benchmark that gives more freedom to do so. Investors
with less discretion may prefer a broader benchmark so
they can make those same active decisions within the
context of their index constraints.
One example of this dynamic can be seen in the choice
of benchmarks for Core Plus fixed-income investors who
have more freedom in their investment choice set. Barclays
Capital publishes two indexes that are commonly used by
US Core Plus investors: the widely used US Aggregate
Index and the US Universal Index. The US Aggregate is
a broad-based investment-grade-only benchmark index
that represents the “Core” part of most broad fixedincome
allocations. The US Universal measures an even
larger universe of dollar-denominated bonds blending the
“Core” US Aggregate Index (86 percent of the US Universal
by market value) with other riskier USD-denominated
bond asset classes such as high-yield, emerging markets,
eurodollar and 144A bonds. Even though the US Universal
may more closely match the Core Plus investment choice
set, a majority of Core Plus managers still use a “Core” US
Aggregate as their benchmark and view the “Plus” as a
strategic yet dynamic out-of-index investment decision to
add alpha to their portfolios.
This also raises an interesting debate about whether outof-index
investments that are allowable as part of a more
broadly defined investment policy are actually true sources
www.journalofindexes.com January / February 2012 45

of alpha if they are not included as part of the index. A
notable example can be seen in the selection of EM hard
currency debt benchmarks. Many investors will use sovereign-only
indexes for their EM portfolios, even though EM
corporates with additional credit risk may be a large and
consistent portion of their portfolios. EM sovereign debt
may represent a majority of the investment choice set, but
if this allocation to corporates is systematic and used by
managers as a consistent way to pick up additional yield, is
any outperformance versus the sovereign-only benchmark
really measuring alpha, or unmeasured beta?
Ultimately, the inclusion of a particular sector or asset
class within a benchmark allows an investor to be either
overweight or underweight that sector, while excluding
an asset class or market segment entirely from a benchmark
may only allow tactical overweights to an issuer
or sector. Either alternative is perfectly reasonable,
especially if an investor feels that excluded markets or
asset classes are not suitable for the portfolio by rule or
by choice. However, if out-of-index assets are a consistent
allocation in an actual portfolio, some may prefer a
benchmark that reflects that broader allocation.
2) Is the portfolio an active or passive product?
Investors who use benchmark indexes are generally
employing one of two strategies: passively tracking an index
return or actively trying to outperform the index. For broad
market indexes or indexes that measure relatively liquid
markets, both active and passive fixed-income managers
generally use the same benchmark for a given fixedincome
asset class. This is because many of the commonly
used market-value-weighted indexes that are used as active
benchmarks can also be efficiently replicated for passive
portfolios without owning every single bond in the index.
In less liquid markets that are more difficult to replicate,
investors may use narrower subsets or tradable
indexes for passive products. An example here would
be the Barclays Capital Emerging Markets Tradable
Inflation Linked (EMTIL) Government Bond Index,
which is sometimes used as a reference index for passive
products instead of the broader benchmark EM
Government Inflation Linked Benchmark Index.
3) Are there explicit portfolio objectives that
must be adhered to?
While some investment policies only specify a broad
asset class allocation with a standard index as its benchmark,
others may have additional constraints in their
portfolio guidelines that need to be reflected in their chosen
index to evaluate manager performance. For example,
an investor managing a defined benefit liability-driven
investment (LDI) portfolio has a specific future liability
stream to fund, generally with a much longer duration
than standard market- value-weighted indexes. Using a
standard index here would clearly not be appropriate for
such an investor-specific objective.
Similarly, an investor may have a specific fixed-income
asset allocation in mind and would like a composite index
to reflect those target policy weights rather than market
value weights of the broad market. Using a standard
market-value-weighted index would also be inappropriate
here if the target allocation is dramatically different.
One other example could be an investor who incorporates
ESG factors in his investment process may have
additional issuer or sector screens in his portfolio that he
would also want reflected in its benchmark index to make
it more suitable for his investment objective.
4) Ultimately, what are the desired risk exposures?
The previous question addressed the presence of
explicit rules or constraints that may exist in an investment
guideline that could influence benchmark selection
and design. A similar consideration must be given to
identifying the investor-specific targeted risk exposures
in a chosen fixed-income index. Market-value-weighted
indexes may represent the broad investment choice
set, but certain investors may have specific market risk
exposures they are seeking for their portfolio and would
like that reflected in their index.
The most common investment characteristic sought
by fixed-income investors is interest rate exposure, either
as a diversifier in a larger multi-asset portfolio allocation
or as an absolute return objective. Some investors
may have a very specific duration view in mind (long or
short compared to the broad market) that they would like
reflected in their benchmark that can be accomplished
through a number of index design alternatives.
Similar benchmark adjustments can be made for other
risk factors such as credit spread risk, currency risk (for multicurrency
bond portfolios), inflation risk and idiosyncratic
risk. Some benchmark design alternatives that manage
these types of exposures are discussed later in this article.
Fixed-Income Index Design Options
Appeal of Market-Value-Weighted Indexes
The most popular fixed-income indexes are market-value-weighted
indexes, which are unmanaged and objective
representations of the broad investment choice set for the
asset class. Weighting by outstanding debt reflects liquidity
and market capacity for the asset class and results in
indexes that are largely replicable by investors managing
against them. They are not optimized indexes or designed
to reflect a specific investment strategy or theme; they
simply measure the returns and risk characteristics of
outstanding debt that meet well-defined and transparent
eligibility criteria. An investor may not prefer the allocations
or risk exposures of the broad market, but will then
use those market views as active investment decisions to
deviate from the market portfolio.
Though complex given the sheer size of the asset class,
market-value-weighted indexes are still easily understood
by a wide variety of index users and are the logical starting
point for passive investors seeking a measure of the asset
class and active investors to formulate investment strategies
to outperform the market. This objectivity, transparency,
universality, market acceptance and coverage of the
46
January / February 2012

asset class are the most common reasons listed by investors
in their preference for this design. It also allows for
better peer group comparisons of asset managers to one
another if they are using the same benchmark.
Alternative Index Designs
Those who prefer a departure from standard marketvalue-weighted
indexes (either inclusion rules or weighting
methodology) often cite a variety of reasons for choosing
an alternative index design. The common thread running
through all of these reasons is that in certain cases,
the index characteristics and risk profile of a market-valueweighted
benchmark may not be the best match for a specific
investor’s portfolio objective or risk tolerance. That is
to be expected, given the size and diversity of the investor
base. In designing or selecting a new index, however, it is
imperative that the properties of a good benchmark are
still present: objectivity, transparency, replicability, etc.
Risks in Switching from Market-Value-Weighted
to Alternative Weighted Indexes
Choosing to move away from widely used index designs
is not without its own risks and costs, especially if one is
switching benchmarks. From a governance and due diligence
perspective, switching benchmarks can take a long
time to get board approval, consultant buy-in, etc.
Another governance question is whether the justification
for using an alternative weight index is really
a benchmark decision or active investment decision.
Many investors and consultants view thematic alternative
weight schemes (ESG weighting, fundamental
weighting, GDP weighting) as subjective investment/
allocation strategies with different risk/return characteristics,
even if they are presented in rules-based form.
Even if an investor agrees with the strategy, the investor
may still prefer to use her existing indexes as benchmarks
and these strategies as a source of alpha generation.
Complicating matters further is that there is less standardization
or agreement on the “right” alternative weight design.
As an example, if you ask a sampling of investors how they
would design a benchmark using fundamentals, you could
get a wide variety of preferences despite supporting the same
general investment theme. Index providers do offer standardized
alternative weight indexes that are transparent and
designed to be used by a wide variety of investors, such as the
Barclays Capital Fiscal Strength Weighted Index family, but
there is far less universal acceptance of a single index design
like there is with market-value-weighted indexes.
Regret risk is another major factor to consider when
choosing alternative index designs. Many investors who
adopt alternatively weighted indexes may, in fact, use
market-value-weighted indexes as a mental benchmark
even after making a switch. If the new index strategy or
design outperforms its market-value-weighted equivalent,
the active decision will be viewed as a success and a suitable
replacement, but if the new index lags or introduces a
considerable amount of incremental risk, an investor may
sometimes reconsider his decision in the short term, even
if the new index was meant to be a long-term solution.
Finally, index replicability is a common concern when
investors choose to deviate from market value weights.
A thematic index may call for substantial increases in
exposure to smaller thinly traded market segments that
may be tougher to replicate. Therefore, liquidity and
capacity become major concerns when choosing to
move away from existing indexes.
A Sample Of Alternative Index Designs
With this framework in place, it is useful to discuss
some examples of common index alternatives
to flagship market-value indexes that are designed to
achieve specific benchmark objectives. The examples
chosen here are available as standard or customized
Barclays Capital Indexes.
• Objective: Measuring a broader investment universe.
Index solutions here tend to expand the scope and coverage
of existing indexes to best measure the full investment
choice set available to investors.
Example: Barclays Capital Global Treasury Universal
Index. We have seen interest from some government
bond investors to broaden their investment choice set and
increase their EM local currency exposure, partially as a
diversification strategy, but also as a source of incremental
portfolio yield. The Global Treasury Universal Index
blends the existing investment-grade Global Treasury
Index with the EM Local Currency Government Index.
• Objective: Measuring a narrower investment universe.
Index solutions of this type introduce additional criteria
to filter down existing market-value-weighted indexes
and are common when investment guidelines are more
restrictive in what can be owned in a portfolio.
Example: Barclays Capital Euro Treasury – Core
Markets Index. With increased focus on sovereign
credit risk within the eurozone, we have seen some
investors choose to exclude riskier or lower-rated
markets such as Greece, Italy, Ireland, etc., from their
benchmark. This index is a version of the standard Euro
Treasury Index, with additional country-exclusion lists
based on credit rating and market size.
• Objective: Matching a specific investment policy
allocation. Index solutions for this objective tend to
be very straightforward blends of existing indexes or
subindexes with weights matching a specific investor’s
policy allocation. As long as any of the subcomponents
exist as stand-alone indexes, almost any allocationbased
index is possible. These are actually the most
common alternative weight indexes.
Example: A U.S. fixed-income manager may decide
that her preferred allocation across sectors/markets
is 50-percent US Agg, 20-percent Global Treasury Ex
US, 10-percent US High-Yield, 10-percent Emerging
Markets, 10-percent Inflation-Linked, and can have
a benchmark created to match that allocation. The
weights are investor specific based on the investor’s
own preferences and target weights.
• Objective: Asset/liability matching. Some portfolios have
www.journalofindexes.com January / February 2012 47

very specific long-duration targets to fund future liability
streams. Even though different LDI indexes may use similar
building blocks (long-duration bond indexes, strips, swaps,
inflation-swaps, etc.), these indexes are investor specific and
tailored to reflect their specific future obligations.
Example: If a client has a projected future liability
stream, he may use a basket of zero coupon nominal and
inflation swaps with different tenors and weights to
match the future expected cash flows to be paid.
• Objective: Managing concentration risk. Index
solutions in this category generally consist of capped
or diversified indexes that apply strict exposure limits
within the benchmark based on issuer weights, sector
weights, country exposure, etc.
Examples: Investors concerned with idiosyncratic risk,
especially prevalent within HY bond portfolios, may use an
index like the Barclays Capital Pan-European High-Yield
2% Issuer Capped Bond Index to limit exposure of any one
issuer to no more than 2 percent of the overall benchmark.
• Objective: Fundamentally themed indexes. Investors
concerned with concentration risk or who hold the
view that “giving more weight to more indebted issuers”
seems counterintuitive 1 sometimes seek more
advanced index options that use other fundamental
factors instead of amount outstanding to determine
allocations within an index. These alternative weights
often buy in to a specific theme or investment thesis
and may change the risk profile of a benchmark.
Example: The Barclays Capital GDP Weighted Bond
Indices family uses GDP as a measure of economy size
to proxy a country’s ability to service its debt and is the
basis for country weights.
Example 2: The Barclays Capital Fiscal Strength
Weighted Indices family uses economic measures of
financial solvency, as well as governance factors, to
tilt and reallocate market-value country weights in
existing benchmark indexes.
• Objective: Increasing liquidity/investability. Investors
sometimes seek more liquid subsets of broad market
indexes or just prefer indexes with less frequent rebalancing,
due to liquidity concerns and high transaction costs.
Tradable baskets are common solutions for investors
seeking more liquid indexes.
Example: The Barclays Capital Tradable EM Local
Currency Bond Index (EMLOCAL) offers diversified
exposure to a basket of more liquid local EM currency
bonds and can be used as a benchmark or access product
for passive exposure.
• Objective: Optimized/risk-weighted indexes. Riskweighted
indexes aim to improve a risk/return trade-off
by deriving weights directly from estimates of market risk
and projected returns. Empirical analysis of historical
returns and covariances underlie their calculations. Riskweighted
indexes have been relatively more popular in
equity markets and, generally, in markets where liquidity
does not represent a major implementation hurdle. In
fixed-income markets, careful combinations of exposures
to underlying risk factors can provide ex-ante efficiency
but are rarely implemented in flagship benchmark indexes;
most of the alternative weight indexes in this category
tend to be bespoke indexes based on market views of
a specific investor and not easily defined with a single
market standard. In fact, the design of an efficient fixedincome
beta is often seen as reflecting the added value of
an individual manager or institution, as opposed to being
a widely available flagship benchmark solution.
• Objective: Managing duration risk. Some investors
may have explicit views on interest rates or specific
target duration (again, perhaps for liability matching)
that is different than the market portfolio. Investors can
choose several index designs to increase/decrease duration
exposure either through rules-based filtering, duration
overlays or duration hedging strategies.
Summary
Are these index alternatives better than market-valueweighted
indexes? For some investors, yes; for others,
definitely not. As an index provider, that is not our evaluation
to make. But by making these types of solutions
available to investors, they have the proper tools at their
disposable to choose or design what they need.
Selecting a benchmark, however, is only the beginning
of implementing and executing a successful passive or
active fixed-income strategy. Even with a chosen benchmark
in place, gaining a deep understanding of index
risk/return characteristics is critical to devising an optimal
strategy to achieve a particular portfolio objective. This
advanced analysis includes: ex-ante tracking error and
tail risk analysis to identify how much risk is being taken
versus a benchmark; scenario testing to understand how
portfolio and index returns are affected in different market
environments; ex-post performance attribution to not
just quantify excess returns versus the benchmark, but
also quantify sources of return and whether they are truly
manager “skill” or maybe just lucky investments.
As we have written before in the Journal of Indexes, a
fundamental principle for successful portfolio management
is to “know thy benchmark,” whether that be a
standard market-value-weighted index or an alternative
index design tailored to a specific portfolio objective.
Endnote
1. This assertion can sometimes be a dangerous oversimplification of a valid portfolio objective (diversification/managing credit risk exposure). The amount of debt outstanding
is not in and of itself a measure of an issuer’s ability to service its debt. Larger entities (government or corporates) tend to issue more debt than smaller issuers,
but in many cases have higher credit ratings, trade at tighter spreads and are not necessarily at greater risk of default based solely on the amount of borrowing they do.
This notion is explored much more deeply in the design of the Barclays Capital Fiscal Strength Weighted Index family that treats market value itself as a fundamental
on liquidity and market capacity.
48
January / February 2012

InsideIndexing
inaugural
conference
BUILDING BETTER BETA 2012
FOR MORE INFORMATION:
www.InsideIndexingConference.com/JOI
From the producers of: InsideETFs InsideETFsEurope InsideCommodities
March 19-20, 2012
5)&3*5;$"3-50/t1)*-"%&-1)*"1"
Designed specifically for INSTITUTIONAL INVESTORS
Featured Speakers:
John C. Bogle
Founder
The Vanguard Group
George F. Will
Pulitzer Prize-Winning
Columnist, Journalist
& Author
Jeremy Siegel
Professor of Finance
University of Pennsylvania
Robert Shiller
Professor of Economics
Yale University
` 'PSTQFBLJOHPQQPSUVOJUJFTQMFBTFDPOUBDU
"MMJTPO+POFTBUPSBKPOFT!*OEFY6OJWFSTFDPN
Presented by:
` 'PSTQPOTPSTIJQPQQPSUVOJUJFTQMFBTFDPOUBDU
'PTUFS8SJHIUBUPSGXSJHIU!*OEFY6OJWFSTFDPN

News
S&P, Dow Jones Indexes
Joining Forces
The S&P 500 and the Dow Jones
industrial average will be united
under the same ownership umbrella,
bringing together two of the biggest
names in indexing in an entity that
will generate more than $400 million
in annual revenue.
The new entity, S&P/Dow Jones
Indices, will be 73 percent owned by
McGraw-Hill, the parent of Standard
& Poor’s, the companies said in a
joint press release issued in early
November. CME Group, which owns
most of Dow Jones indexes, will hold
a 24.4 percent stake in the venture,
while Dow Jones & Co. Inc. will hold a
2.6 percent stake. The transaction has
approval of both companies’ boards.
The two widely used benchmarks in
the world of investing having the same
owners potentially means they would
have a clear edge over other indexingindustry
competitors, notably MSCI,
which was outbid by CME Group last
year in its attempt to acquire Dow
Jones Indexes from Dow Jones & Co.
Inc. CME Group now owns 90 percent
of Dow Jones Indexes, while Dow
Jones & Co., the News Corp. unit, currently
holds a 10 percent stake.
The press release said that the joint
venture would be up and running
within the first six months of 2012.
The new indexing company will
become part of the new McGraw-Hill
Markets, which will come out of the separation
of McGraw-Hill into two public
companies. McGraw-Hill announced
plans for that separation on Sept. 12.
The companies said that under
their agreement, the S&P/Dow Jones
Indices joint venture will enter into a
new license agreement under which
the CME Group will pay S&P Indices
a share of profits from the CME Group
equity products lineup.
That new license agreement also
expands products covered under the
license to include swaps, and extends
CME Group’s currently exclusive rights
to E-mini futures and other indexed
S&P indexed futures. That exclusive
agreement is in place until the end of
2017, the companies said.
Alexander Matturri, executive
managing director of S&P Indices,
will be the chief executive officer of
S&P/Dow Jones Indices.
Hope For Housing?
U.S. home prices saw another
minor uptick in August, fueling a
“modest glimmer of hope” about a
potential market recovery, according
to the S&P/Case-Shiller report on U.S.
residential real estate market.
The August S&P/Case-Shiller
report, released in October, shows that
both the 10-city and 20-city composites
were up 0.2 percent that month,
respectively, from July levels, but its
continued improvement in year-overyear
readings has analysts optimistic.
Still, the housing market isn’t out of
the woods yet, which was at the center
of the 2008 credit crisis, and has yet to
stage a sustainable recovery deemed
crucial to spur sustainable economic
growth. Earlier this year, the market
retested 2009 lows and has since stabilized
just above those levels, but
hasn’t moved much more.
Compared with last year, however,
the hole seems to be getting shallower.
While prices generally remained below
year-earlier levels in August, the latest
report shows year-over-year comparisons
improving in 16 of the 20 cities surveyed.
The exceptions are Los Angeles,
Miami, Atlanta and Las Vegas.
That said, both the 10-city and
20-city composites are still more than
30 percent off their peak readings back
in 2006, with national home prices
still hovering around their mid-2003
levels. Also, the monthly statistics
showed weakness, with half the cities
surveyed showing price drops in
August relative to July.
SSgA & PowerShares
Clash Over KBW Products
On Oct. 24, State Street Global
Advisors changed the indexes on
five of its ETFs from benchmarks
provided by KBW to a quintet from
Standard & Poor’s in a shift that belied
a bit of intrigue involving Invesco
PowerShares.
Indeed, four of the KBW indexes
SSgA dropped were adopted by
four new ETFs PowerShares rolled
out Nov. 1. One industry source told
IndexUniverse that SSgA sped up the
transition to the new indexes before
the launch date of the PowerShares
funds—a move PowerShares hoped to
block, possibly with legal action.
Officials at both companies declined
to comment on the timing and legal
issues surrounding the index changes
and the PowerShares fund launches.
An SSgA official said his Bostonbased
firm made the changes to line
up the indexes on the five financerelated
ETFs with its other sector
funds, all of which are based on S&P
indexes. Meanwhile, Wheaton, Ill.-
based PowerShares was interested in
expanding its relationship with Keefe,
Bruyette & Woods and taking advantage
of its KBW lineup of financespecific
indexes. PowerShares has
a few products using KBW indexes,
including the $21.7 million KBW High
Dividend Yield Financial Portfolio
(NYSE Arca: KBWD).
The SSgA funds retained their tickers,
but were renamed, with “S&P”
replacing “KBW”:
• SPDR S&P Bank ETF (NYSE Arca: KBE)
• SPDR S&P Capital Markets ETF
(NYSE Arca: KCE)
• SPDR S&P Insurance ETF (NYSE
Arca: KIE)
50
January / February 2012

• SPDR S&P Mortgage Finance ETF
(NYSE Arca: KME)
• SPDR S&P Regional Banking ETF
(NYSE Arca: KRE)
The four new PowerShares ETFs
include:
• PowerShares KBW Bank Portfolio
(NYSE Arca: KBWB), which is based
on the KBW Bank Index
• PowerShares Regional Banking
Portfolio (NYSE Arca: KBWR), which
is based on the KBW Regional
Banking Index
• PowerShares Capital Markets
Portfolio (NYSE Arca: KBWC),
which is based on the KBW Capital
Markets Index
• PowerShares Insurance Portfolio
(NYSE Arca: KBWI), which is based
on the KBW Insurance Portfolio Index
ETFs Under Scrutiny
In Senate Hearing
The fast-growing world of exchangetraded
funds came under close scrutiny
on Oct. 19 at a Senate Banking, Housing,
and Urban Affairs subcommittee hearing
chaired by Jack Reed (D-RI).
Noel Archard, a managing director
at the world’s biggest ETF firm,
iShares, and Harold Bradley, chief
investment officer at the nonprofit
Kauffman Foundation, were
among those testifying at the U.S.
Senate Banking subcommittee
hearing, “Market Microstructure:
Examination of Exchange-Traded
Funds (ETFs).”
Archard and the other two panelists—the
Nasdaq exchange’s
Eric Noll, and Eileen Rominger,
the director of the Securities and
Exchange Commission’s Investment
Management division—generally
agreed that ETFs aren’t to blame for
all the volatility in the markets in
the past few months. However, the
Kauffman Foundation’s Bradley, coauthor
of two recent research reports
taking a critical view of ETFs, argued
that ETFs represent a risk to the functioning
of modern equity markets and
called for a broad SEC inquiry into
their impact on markets, suggesting
that ETFs are “dangerous.”
It was clear the other panelists
didn’t share Bradley’s concern—or at
least the extent of it. While calling for
greater disclosure and transparency,
they emphasized that ETFs are among
the most cost-efficient and tax-efficient
investment vehicles.
The SEC’s Rominger, for example,
said that while the commission remains
committed to ensuring investors are
properly protected, its inquiries so far
haven’t led the SEC to the conclusion
that ETFs are roiling the markets.
Noll, who is in charge of transaction
services at Nasdaq, was plain in
noting that all the recent volatility has
a lot more to do with real uncertainty
than with ETFs.
There was more of a consensus
between Bradley and the other panelists
on the need for regulation of the relatively
small piece of the ETP market that
involves leveraged and inverse ETFs.
All the panelists seemed to agree
that there’s a need to protect the
interest of investors who might not
understand ETFs, especially those
using derivatives to execute their
investment strategies.
Archard from iShares testified that
leveraged and inverse ETPs should
not even be labeled ETFs, and he
advocated a new regulatory standard
that establishes clear guidelines for
delineating derivatives such as leveraged
ETFs from plain-vanilla ones.
Archard also called for regulations
that require disclosures of any
On Oct. 24, State Street Global Advisors
changed the indexes on five of its ETFs.
www.journalofindexes.com January / February 2012 51

News
costs or fees that might affect investors’
holdings and returns.
Bradley, meanwhile, specifically
called for much more stringent regulation
for market makers, arguing they
were exempt from too many rules.
INDEXING DEVELOPMENTS
Global Russell Indexes
Target Stability
In October, Russell Investments
launched global versions of the U.S.-
focused “stability” factor-based indexes
it debuted in February that are
designed to separate companies based
on whether they are likely to thrive in
a healthy economy or reliably weather
the storm of a downturn. The benchmarks
include the Russell Global
Defensive and Dynamic Indexes.
The cap-weighted benchmarks
blend various fundamental factors—
such as leverage, returns on assets and
earnings at a company level—with
market volatility in an effort to determine
how susceptible a company is to
a changing economic environment.
After being ranked, the chosen
equities are then split into halves—one
more stable and the other riskier. Those
halves are then grouped into so-called
Defensive indexes on the one hand
and Dynamic indexes on the other, the
company said in a press release.
In the global version, stocks are
ranked by sector and style across
regions rather than by country in an
effort to better reflect the way investors
approach global market exposure,
the company said in the release.
FTSE Expands IPO Indexes
In late September, FTSE announced
it would be adding another 53 indexes to
its IPO index series. Newly rechristened
the FTSE Renaissance Global IPO Index
Series, it covers 518 stocks from developed
and emerging markets around the
world, and can be broken out into country
and regional benchmarks.
According to the press release, customized
sector-based indexes derived
from FTSE’s Industry Classification
Benchmark are also available, in addition
to other types of bespoke indexes.
Eligible components of FTSE’s IPO
indexes are typically added to the series
when they begin trading and remain
for roughly two years, the release said.
MSCI Launches Blue-Chip ‘EM 50’
MSCI launched a version of its popular
MSCI Emerging Market Index in
early October that cherry-picks 50 of
the broader index’s largest components,
thereby alleviating some of the
replication challenges associated with
smaller, less liquid constituents.
The MSCI EM 50 Index remains
highly correlated to the flagship
index, and so is very tradable, the
company said in a press release.
In the interests of liquidity, the
methodology screens out markets with
weightings of less than 3 percent in the
MSCI Emerging Markets Index. Also,
stocks from Brazil, India, Mexico and
Russia are excluded from the selection
universe due to investability issues;
instead, those markets are represented
through depositary receipts. Other
screens are applied to maximize liquidity
and investability for foreign investors,
and the largest 50 securities are selected
from the resulting selection pool based
on free-float market capitalization,
according to the index methodology.
New S&P Indexes Cover Asian
Commodities Producers
In October, S&P introduced two
indexes covering commodities producers
in Asia, a press release said.
The S&P Asia Commodity Producers
Oil, Gas and Coal Index and the S&P Asia
Commodity Producers Agribusiness
Index include securities from Asian
markets, with the exception of China
and India, and certain types of securities
listed on the Thailand market. However,
Hong Kong-listed securities for Chinese
companies are eligible.
The indexes pull their components
from the S&P Pan Asia BMI, according
to the press release. The energy-commodities
index includes stocks from the
oil & gas exploration & production and
the coal & consumable fuels categories
of the Global Industry Classification
Standard. The agribusiness index covers
stocks from the agricultural products,
the construction & farm machinery &
heavy trucks, the fertilizers & agricultural
chemicals and the packaged foods
& meats designations.
The indexes are weighted by modified-market
capitalization, and cap
the weightings of China and Hong
Kong listings at a total of 40 percent,
the press release said.
Markit Leveraged
Loan Index Debuts
Markit unveiled its Markit iBoxx
USD Liquid Leveraged Loan Index in
mid-October.
Leveraged loans are typically made
to heavily indebted borrowers and carry
a higher rate of interest due to their
greater likelihood of default. The Markit
index draws its 100 components from
a universe of 3,000 such loans denominated
in U.S. dollars. They are selected
based on a variety of liquidity factors,
with prices based on multiple independent
sources, the press release said.
The company already had a
broad leveraged loan index with 850
components, the Markit iBoxx USD
Leveraged Loan Index, according to
the press release.
Russell, Chi-X Launch
European Indexes
Pan-European equities exchange
Chi-X Europe and Russell Indexes
rolled out the Chi-X Europe Russell
Index (CHERI) Series in early October.
The index family covers Europe and
draws its components from the Russell
Global 1000 Index, a press release said.
The four new indexes include the
Chi-X Europe Russell PanEurope
Index, Chi-X Europe Russell Eurozone
Index, Chi-X Europe Russell PanEurope
60 Index and Chi-X Russell Europe
Eurozone 40 Index—two broader benchmarks
and two blue-chip indexes.
The index family combines Russell’s
index methodology with Chi-X’s pricing
data, the press release said. The
statement also said that the indexes
were constructed to be investable,
tradable and closely correlated to
other European indexes.
52
January / February 2012

FTSE Adds EM To
DBI Series Lineup
FTSE kicked off November with the
announcement that it was adding an
emerging markets index to its FTSE
Diversification Based Investing Index
family. The new benchmark joins the
FTSE DBI Developed Index.
The top components in the FTSE
DBI All Emerging Index, as of Sept.
30, 2011, were Chile’s Empresas
Copec SA, Morocco’s Maroc Telecom
and the Philippines’ Manila Electric,
according to a fact sheet.
The indexes are designed to minimize
concentration risk by diversifying
across sectors and countries based on
correlation data. The family of indexes
was originally launched in September
2010 and was developed with QS
Investors LLC, the fact sheet said.
S&P Debuts LatAm
Infrastructure Benchmark
In late September, S&P rolled out
the S&P Latin America Infrastructure
Index, which targets Latin American
infrastructure companies in the areas of
energy, transportation, telecommunications
and utilities, a press release said.
According to a fact sheet, companies
domiciled in Argentina, Brazil, Chile,
Colombia, Mexico, Panama and Peru
with market capitalizations of at least
$200 million and three-month average
daily trading value of more than
$1 million are eligible for inclusion in
the index. The methodology allocates a
10 percent weighting to energy-related
stocks, with the transportation, telecom
and utilities categories each weighted
at 30 percent of the index.
At launch, the index had a total of
22 components, the fact sheet said.
Dow Jones Indexes Launches
Europe, Asia Dows
Dow Jones Indexes in mid-October
debuted European and Asian versions
of the Dow Jones industrial average.
Like the DJIA, the new indexes
each track 30 blue-chip stocks selected
subjectively from their target
markets, the press release said. The
Averages committee, which is responsible
for choosing the components of
the DJIA and the 150-stock Global
Dow, selects the components of the
Europe Dow and the Asia Dow.
The European index includes such
names as BP Plc, Allianz SE and Total
SA, the press release said. Meanwhile,
the Asian index includes such names
as China Mobile Ltd., BHP Billiton
Ltd. and Toyota Motor Corp.
The press release noted that while
the DJIA is price weighted, the new
indexes are equal weighted.
New Stoxx Indexes
Exclude Financials
Stoxx Limited, in an apparent
acknowledgment of investor concerns
regarding Europe’s financial sector,
launched a group of 12 indexes
that exclude either the Industry
Classification Benchmark’s financials
industry or banks supersector, a mid-
October press release said.
The new indexes are based on the
same methodology as the older benchmarks
from which they are derived,
according to the press release. The list
includes modified versions of the following
benchmarks:
• Euro Stoxx
• Euro Stoxx Large
• Stoxx Europe Large
• Stoxx Europe 600
• Euro Stoxx 50
• Stoxx Europe 50
BarCap Unveils Dim Sum Index
Barclays Capital rolled out the
Barclays Capital Offshore Renminbi
(CNH) Bond Index in early October, a
press release said. The new benchmark
targets what is popularly known as the
“dim sum” market—bonds issued outside
of mainland China but denominated
in China’s local currency.
The index is a stand-alone, the
press release noted, adding that the
new benchmark joins BarCap’s other
indexes targeting China and Asia,
such as the Barclays Capital China
Aggregate Index—also denominated
in renminbi and targeting companies
in mainland China—and the Barclays
Capital Asian Pacific Aggregate Index.
DJ, BBVA Team Up
On Eagles Index
Dow Jones Indexes and Spainbased
bank BBVA Group rolled out
two indexes in late September that
focus on companies located in Brazil,
China, Egypt, India, Indonesia, Mexico,
Russia, South Korea, Taiwan and
Turkey. The Dow Jones BBVA Eagles
Index and the Dow Jones BBVA Eagles
Optimized Index, whose constituent
countries will be reviewed annually,
are designed to measure the performance
of 50 companies in “emerging
and growth-leading economies”—
hence the Eagles moniker.
They represent only the largest
companies in the targeted countries
and stocks must pass liquidity
screens to be eligible for the index, the
index’s creators said. Companies are
selected and weighted to reflect the
GDP-based Eagles country weightings.
Company weights are restricted
based on their liquidity. The indexes
include only U.S., EU and Hong Kong
listings, with the aim of enhancing
their liquidity and easing their replication
by foreign investment firms.
2012 Weights For
DJ-UBSCI Announced
In October, Dow Jones Indexes
and UBS announced the 2012 target
weightings for the Dow Jones-UBS
Commodity Index in a press release.
The biggest change is the inclusion
of the Brent futures contract in the
index. Previously, crude oil had been
represented solely by the WTI futures
contract, which had a 2011 target
weight of approximately 14.7 percent.
In 2012, crude oil will be represented
by both contracts, the press release
said, with the WTI contract receiving a
target weighting of roughly 9.7 percent
and the Brent contract’s weighting set
at approximately 5.3 percent.
None of the other commodities
contracts in the index saw
meaningful changes to their target
weightings from 2011.
The changes to the index will be
effective as of January 2012, the press
release said.
www.journalofindexes.com January / February 2012 53

News
AlphaClone Debuts
Index, Plans ETF
AlphaClone, a firm that says it
enables investors to “clone” what the
top investment managers are doing,
unveiled a new “alternative alpha”
index in October that seeks to generate
the same risk/return benefits of hedge
funds. It also said the index is likely to
become the basis of an ETF the company
hopes to roll out in 2012.
The rules-based AlphaClone Hedge
Fund Long/Short Index tracks the performance
of U.S. equities through a proprietary
selection process that rebalances
positions quarterly. The strategy focuses
on mitigating downside volatility and
reducing overall market correlations, the
company said in a press release.
AlphaClone didn’t provide any
additional details on the ETF it said it
has in the works.
Nasdaq, PC-Bond Offer
Treasury Indexes
A November announcement from
Nasdaq OMX indicated that the exchange
had teamed up with bond index provider
PC-Bond to create U.S. Treasury indexes.
The RBC Insight Total Return U.S.
Treasury (TRUST) Indexes rebalance
daily, according to the press release,
which also noted that most bond indexes
rebalance on a monthly basis. The family
includes 22 indexes covering different
maturities, as well as subindexes that
exclude “stripped” securities, the Nasdaq
website said.
A back history has been calculated
from Dec. 31, 1998, on, the press
release noted.
AROUND THE WORLD OF ETFs
First-Ever Copper ETF Debuts
United States Commodity Funds
launched the first U.S. ETF to invest in
copper futures when it rolled out its
United States Copper Index Fund (NYSE
Arca: CPER) in November. The product
carries an expense ratio of 0.95 percent.
The fund features an unusual
methodology designed to alleviate the
effects of contango, generally investing
in two nearer-month contracts
displaying the greatest degree of backwardation,
but adding a third contract
expiring significantly further out on
the futures curve when the market
slips out of backwardation.
There were already two ETNs available
that covered the copper futures
market, but CPER is the first product to
do so utilizing the ETF structure.
New iShares Target Low Volatility
In late October, BlackRock’s iShares
unit debuted four international ETFs
targeting low-volatility stocks.
The four new funds use MSCI indexes
that are effectively subsets of four
broad-based MSCI indexes that have
been screened to select the stocks that
bounce around in price a bit less than
the rest of the components in the broad
indexes. Each of the minimum-volatility
ETFs applies a rules-based methodology
to determine the weight of the securities
based on risk, iShares said in regulatory
paperwork it filed earlier this year.
The four funds and their expense
ratios are:
• iShares MSCI USA Minimum
Volatility Index Fund (NYSE Arca:
USMV), 0.15 percent
• iShares MSCI EAFE Minimum
Volatility Index Fund (NYSE Arca:
EFAV), 0.20 percent
• iShares MSCI All Country World
Minimum Volatility Index Fund
(NYSE Arca: ACWV), 0.35 percent
• iShares MSCI Emerging Markets
Minimum Volatility Index Fund
(NYSE Arca: EEMV), 0.25 percent
Global X Launches
Social Media Fund
Global X, a firm known for its niche
sector ETFs, debuted a social media
ETF in mid-November.
The Global X Social Media Index
ETF (Nasdaq GM: SOCL) is the first
portfolio allocated entirely to social
media, tracking the 25-security freefloat,
market-capitalization-weighted
Solactive Social Media Index. It comprises
companies involved with social
networking, file sharing and other Webbased
applications. SOCL comes with a
net expense ratio of 0.65 percent.
The fund’s top holdings at launch
included Tencent Holdings, Sina
Corp, DeNA Co. and NetEase.com.
China represented about 37 percent
of the index, followed by the U.S. at 26
percent and Japan at 20 percent.
Pimco Rolls Out Country-
Specific Bond ETFs
During the first half of November,
Pimco launched a series of countryspecific,
fixed-income ETFs. Each of
the funds targets local-currency, investment-grade
debt issued by a developed
market that has weathered the recent
economic downturn fairly well in relation
to other developed markets.
The funds include the following:
• Pimco Australia Bond Index Fund
(NYSE Arca: AUD)
• Pimco Canada Bond Index Fund
(NYSE Arca: CAD)
• Pimco Germany Bond Index Fund
(NYSE Arca: BUND)
Each charges an expense ratio of
0.45 percent.
Vanguard Plans Int’l Bond Funds
Vanguard kicked off November
with a bit of a bombshell—a filing for
two international bond index mutual
funds that included ETF share classes
for each. International fixed income
has long been a glaring omission from
the fund provider’s otherwise com-
54 January / February 2012

prehensive lineup of index funds.
The two funds and the estimated
expense ratios of the ETF versions are:
• Vanguard Total International Bond
Index Fund, 0.30 percent
• Vanguard Emerging Markets
Government Bond Index Fund,
0.35 percent
Both will track Barclays Capital
indexes. The international bond fund
will be hedged, meaning that whatever
currency exposure a given bond
has will be neutralized by the fund’s
hedging strategy, Vanguard said in
a press release on the two proposed
funds. The emerging markets fund will
meanwhile own dollar-denominated
credits, according to the filing.
Guggenheim Launches
Renminbi Fund
In early October, Guggenheim
Funds added to its lineup of
CurrencyShares funds by launching
the CurrencyShares Chinese Renminbi
Trust (NYSE Arca: FXCH), which is
designed to reflect the value of China’s
currency, the renminbi, in dollars.
Like the rest of the currency shares,
the fund is structured as a grantor trust
and simply holds the targeted currency.
FXCH comes with an expense ratio
of 0.40 percent.
The new fund’s primary competitor
is the WisdomTree Dreyfus Chinese
Yuan Fund (NYSE Arca: CYB), which
is more expensive, carrying a 0.45 percent
expense ratio.
Russell Debuts Int’l Factor ETFs
Russell Investments added to its
growing lineup of factor-based ETFs
in early November with the launch
of three international funds. The new
funds, like their U.S.-focused counterparts,
use indexes that single out
stocks with low beta, low volatility and
high momentum.
The three funds, which all have
net annual net expense ratios of 0.25
percent, are:
• Russell Developed ex-U.S. Low Beta
ETF (NYSE Arca: XLBT)
• Russell Developed ex-U.S. Low
Volatility ETF (NYSE Arca: XLVO)
• Russell Development ex-U.S. High
Momentum ETF (NYSE Arca: XHMO)
The underlying indexes are all subsets
of the Russell Developed ex-U.S.
Large Cap Index.
BACK TO THE FUTURES
CME Volume Up In October
A CME Group press release said
that its volume for October 2011 was
up 9 percent from October 2010,
averaging 12.4 million contracts per
day. Equity-index contracts were the
best-performing segment, with a 29
percent year-over-year increase in
average daily volume. Meanwhile,
interest-rate contracts were the worstperforming
segment, with average
daily volume up just 2 percent.
Monthly volume for the exchange
group’s most actively traded index
futures contract, the e-mini S&P 500,
was up 31.3 percent year-over-year to 55
million contracts. Trading in the e-mini
Nasdaq-100 contract was up just 9 percent
to 6.8 million contracts, and the
e-mini $5 Dow contract’s volume was
at 2.8 million contracts for the month, a
year-over-year increase of 4.6 percent.
KNOW YOUR OPTIONS
CBOE Sees Options
Volume Increase
CBOE Holdings saw its options volume
increase 16 percent year-overyear
in October to an average daily
volume of 5 million contracts. Almost
40 percent of that daily volume was
ETF- or index-based options.
While index options saw their average
daily volume increase 35 percent
year-over-year to 1.4 million contracts
in October, ETF options were
up 49 percent to 1.5 million contracts.
The most actively traded of those
contracts included the options on the
S&P 500 Index, SPDR S&P 500 ETF,
VIX, iShares Russell 2000 Index Fund
(NYSE Arca: IWM) and PowerShares
QQQ (Nasdaq GM: QQQ).
FROM THE EXCHANGES
CBSX To Acquire NSX
The CBOE Stock Exchange (CBSX),
the all-electronic exchange created by
the Chicago Board Options Exchange
(CBOE) and four market-maker partners
in 2007, is acquiring the National
Stock Exchange (NSX), according
to an announcement at the end of
September. Terms weren’t disclosed.
The transaction, which is pending
regulatory approval, would bring data
systems and business operations from
both exchanges together into one
platform. But the CBOE hopes to keep
the New Jersey-based National Stock
Exchange as a separate unit, it said
in a press release. The all-electronic
National Stock Exchange is currently
owned by several broker-dealers.
The press release made clear that
the plan is friendly and mutual.
ON THE MOVE
CBOE Names New President/COO
CBOE Holdings kicked off November
with the announcement that
President and COO Edward Joyce
was resigning due to medical reasons.
Edward Tilly, the company’s executive
vice chairman, was named as his successor,
according to the press release.
Tilly became executive vice chairman
in August 2006. Previously, he was
the CBOE member vice chairman, the
press release said.
Joyce became president and COO of
CBOE Holdings in June 2000, according
to the press release. Prior to that,
he was the executive vice president of
business development.
McIsaac Made Head Of
Vanguard Institutional Group
Pensions & Investments reported in
November that Vanguard had placed
Christopher McIsaac at the head of its
Institutional Investor group, succeeding
Greg Barton. McIsaac also was to takeover
leadership of Vanguard’s full-service
institutional business from Barbara
Fallon-Walsh, the article noted.
McIsaac is a Vanguard principal who
previously led the company’s Portfolio
Review group, P&I said. That group is
now headed by Sean Hagerty.
Barton and Fallon-Walsh were
scheduled to retire at the end of 2011,
the article said.
www.journalofindexes.com January / February 2012
55

Global Index Data
56
January / February 2012

Index Funds
www.journalofindexes.com January / February 2012
57

Morningstar U.S. Style Overview Jan. 1 - Oct. 31, 2011
Source: Morningstar. Data as of Oct. 31, 2011
Source: Morningstar. Data as of 2/29/08
58
January / February 2012

Dow Jones U.S. Industry Review
www.journalofindexes.com January / February 2012
59

Exchange-Traded Funds Corner
60
January / February 2012

Shepherd continued from page 23
Figure 3
Fundamentally Weighted Broad-Based Bond Returns Vs. Cap-Weighting, January 1998–June 2011
Portfolio
Return
Standard
Deviation
Sharpe
Ratio
Average
Credit Rating
Duration
Excess
Return
Fundamentally Weighted
Broad-Based Index
Cap-Weighted
Aggregate Index
Barclays Capital Global
Aggregate Index
7.73% 7.23% 0.69 AA2 5.62 1.61%
6.62% 6.62% 0.59 AA2/AA3 5.73 0.50%
6.12% 6.04% 0.56 – – –
Source: Research Affiliates, based on data from Bloomberg
Figure 4
Fundamentally Weighted Vs. Cap-Weighted Aggregate Sector Allocations, June 2011
Portfolio
EM Sovereign
Local
EM Sovereign
USD
Developed
Sovereign
EM Corporate
Global
Yield
Global Inv
Grade
Fundamentally Weighted
Broad-Based Index
Cap-Weighted
Aggregate Index
2.7% 1.8% 65.6% 0.7% 4.3% 26.3%
2.7% 1.1% 67.6% 1.0% 4.1% 23.4%
Source: Research Affiliates, based on data from Bloomberg
component pieces compose over 90 percent of the index
weight on average, the aggregate portfolio alpha is significantly
higher than the individual alpha for these component
pieces. This additional alpha arises as the portfolio
contratrades against growing or shrinking credit spreads.
The value of a systematic rebalancing across asset classes
should not be surprising to anyone familiar with the basic
principles of asset allocation; yet the construction of typical
cap-weighted aggregate bond indexes entirely forgoes
this easily achievable return advantage. In fact, it does
the exact opposite: The return drag on the cap-weighted
aggregate index grows larger as it systematically increases
exposure to Treasurys when they become more expensive,
and moves out of credits when they trade more cheaply
and offer a more compelling risk/return trade-off.
The fundamentally weighted broad-based portfolio
not only avoids this return drag, but generates outperformance
through systematic rebalancing across bond
sectors. As a result, the index as a whole is indeed more
than the sum of its parts.
References
Arnott, Robert D., Jason C. Hsu, Feifei Li, and Shane D. Shepherd. 2010. “Valuation-Indifferent Weighting for Bonds.” Journal of Portfolio Management, vol. 36, No. 3
(Spring):117-130.
Arnott, Robert D., Jason C. Hsu, and Philip Moore. 2005. “Fundamental Indexation.” Financial Analysts Journal, vol. 61, No. 2 (March/April):83-99.
Enderle, Francis J., Brad Pope, and Laurence B. Siegel. 2003. “Broad-Capitalization Indexes of the U.S. Equity Market.” Journal of Investing, vol. 12, No. 1 (Spring):11-22.
Hsu, Jason C. 2006. “Cap-Weighted Portfolios are Sub-Optimal Portfolios.” Journal of Investment Management, (Third Quarter): 1-10.
Shepherd, Shane D. 2011. “Fundamental Index Fixed-Income Performance: It’s All in the Price.” Journal of Index Investing, vol. 1, No. 4 (Spring):75-80.
Endnotes
1. Our broad-based fundamentally weighted index is compared to both the broad BarCap Global Aggregate as well as our internally calculated cap-weighted benchmark
because it is difficult to generate fundamental factors needed to generate fundamental weights for some sectors of the BarCap Global Aggregate (e.g., securitized debt).
2. In fact, it may be helpful to think of inflation risk as a form of default risk. Even if the face value is paid back in nominal dollars, the purchasing power of those future dollars
will be eroded by some unknown amount; thus, a percentage of the bond has “defaulted” in real terms.
3. Some privately held companies issue publicly traded debt but do not always report publicly available financial information, thus making it difficult to include the securities
in the fundamentally weighted index.
4. We compute separate returns for the book value, sales, cash flow and dividends for the corporate indexes; and for GDP, population, land area and energy consumption for
the sovereign indexes. These results are not reported here for the sake of brevity but are available upon request.
5. The BarCap Global Aggregate does not break down the sectors into quite the same components, but its allocation is approximately 67 percent Treasurys or governmentrelated
debt (dominated by the developed-world governments); 16 percent corporate debt; and 17 percent securitized issues. Except for the presence of the securitized debt,
the allocations do not differ dramatically from our cap-weighted portfolio.
www.journalofindexes.com
January / February 2012
61

Benzschawel continued from page 31
Endnotes
1
The WGBI is described in detail in the Citigroup Global Fixed-Income Index Catalog [Citigroup Index, 2011].
2
A derivation can be found in Choudhry [2006].
3
See Figures 18 and 19 of Benzschawel and Lee [2011].
4
A detailed description of the method appears in Benzschawel and Lee [2011].
5
We wished to use as short a spread history as possible in order to generate PDs for the widest universe of credits and found a six- to nine-month series of daily spreads sufficient.
However, the diffusion of credit spreads over time cannot be Gaussian; credit spreads are bounded by the risk-free rate and by the price difference between par and
bonds’ recovery values in default. If so, expected spreads for longer-duration bonds based on short-duration volatility estimates will be greater than observed values, owing
to their lesser realized volatility (see Benzschawel and Lee [2011]). Because the structure of long-term spread volatility is not well documented, our provisional solution to
this problem is to convert bond spreads of maturities greater than one year to their one-year equivalent spreads.
6
Although structural and hybrid methods for estimating PDs have problems (see Keenan, Sobehart and Benzschawel [2003], those were not found to be limitations to their
use for estimating the credit risk premium. In particular, although individual PDs from the HPD model may be suspect in some cases, PDs over a large sample of firms may
be assumed to be correct on average.
7
Rating category assignments are intended to be indicative only.
8
Many sovereigns have not issued debt in USD, so their spreads to U.S. Treasurys are not available. To generate PDs for those issuers, bond yields in foreign currencies are
converted to equivalent USD-denominated yields by subtracting the difference between the foreign one-year swap rate and the USD one-year swap rate from the foreign
bond yields. The U.S. Treasury yield is then subtracted from the U.S. equivalent yield to generate the series of spreads to U.S. Treasurys that can then be used along with the
issuers’ USD-denominated bonds to calculate market-implied PDs.
9
Of course, assuming a 40 percent recovery value for all sovereigns, or all corporates for that matter, is questionable. Hence, this PD versus spread analysis must be viewed
as speculative.
10
In this case, average rating refers to assigning single-integer distances between rating categories and then averaging. For example, AAA = 1, AA+ = 2,…, D = 22.
11
PDs in the linear algorithm are expressed in percent. However, for the logarithmic PDs, we converted from percent to basis points (bp), where 100 bp = 1 percent, to avoid
negative PDs for firms with market-implied PDs less than 1 percent.
12
The antilog of x in base 10
is 10 x , such that the antilog of 2 is 100.
13
The sum of the contributions for the multiplicative scheme in Equation 10 can be greater than 100 percent. Accordingly, the contributions, w i
, for each obligor were divided
by the sum of all contributions so that the resulting contributions sum to 100 percent.
References
Benzschawel, T. and Lee, M. “Market-Implied Default Probabilities: Update,” Citi, September 9, 2011.
Choudhry, M. “The Credit Default Swap Basis,” Bloomberg Press, New York, 2006.
Citigroup Index, “Citigroup Global Fixed-Income Index Catalog—2011 Edition,” Citi, February 8, 2011.
Keenan, S., Sobehart, J. and Benzschawel, T. “The Debt and Equity Linkage and the Valuation of Credit Derivatives,” in “Credit Derivatives: The Definitive Guide,” J. Gregory
(Ed.), Risk Books, London, pp. 67-90, 2003.
Markowitz, H. “Portfolio Selection,” Yale University Press, 1959.
Nelson, C.R. and Siegel, A.F., “Parsimonious Modelling of Yield Curves,” Journal of Business 60 (4), pp. 473-89, 1987.
Sobehart, J. and Keenan, S. “Hybrid Contingent Claims Models: A Practical Approach to Modeling Default Risk,” in “Credit Ratings: Methodology, Rationale and Default Risk,”
M. Ong (ed.), Risk Books, 2002
Sobehart, J. and Keenan, S. “Hybrid Probability of Default Models: A Practical Approach to Modeling Default Risk,” Quantitative Credit Analyst 3, pp. 5-29, 2003.
Wirz, M. “Raters Fail to See Defaults Coming,” Wall Street Journal, pp. 1-2, August 12, 2011.
Why advertise in the Journal of Indexes?
JOURNAL OF INDEXES ADVERTISING INFORMATION AT WWW.JOURNALOFINDEXES.COM/ADVERTISE
IndexUniverse LLC, 353 Sacramento St., Suite 1520, San Francisco, CA 94111 • Advertising and Reprints Inquiries: 415.659.9004
62 January / February 2012

Wiggins continued from page 39
would offer more diversified exposure. By selecting an index
that’s more than three-quarters government bonds, investors
may be taking on more risk than they’re aware of, as
interest rate risk accounts for 98 percent of the total return
behavior of government bonds. Bond prices are affected by
supply and demand, just like all securities, and the growing
issuance of government debt seems like a certainty, while
the same can’t be said of the corporate market.
Beyond The Agg
According to ICI data, as bond funds have experienced
massive inflows in the area of hundreds of billions of dollars
in the last few years, money market funds have experienced
even more massive outflows—to the tune of more than $1
trillion. While the money flowing out of money markets no
doubt went to a number of sources (and much of the bond
inflows were no doubt shifted from equities), it doesn’t seem
unreasonable that many investors could have traded the nextto-nothing
return on their money market fund for a 3 percent
bond fund. Yet today, well under 20 percent of Barclays
Aggregate assets are allocated to investment-grade corporate
bonds—a corner of the fixed-income market where most of
the holdings have a coupon of 5 percent or more—offering
one of the few means of achieving attractive current returns
in a low-yield environment. Can interest rates decline from
historic lows? Nobody knows for sure, but the higher yield
of corporates would not only cushion a future rise in interest
rates, it would carry less systematic or market risk. Even in
2008 and 2009, only a handful of investment-grade corporate
bond issuers defaulted or entered bankruptcy.
In fact, performance in recent months has come from segments
that aren’t well represented in the Aggregate index. In
2010, corporate bonds had an excess return of 2.3 percent,
as profitability increased and cash levels grew to the highest
percentage of total corporate assets since 1959. Commercial
mortgage-backed securities, about 2 percent of the Barclays
Agg at the end of November 2011, delivered an even stronger
performance—an excess return of nearly 15 percent—as
many investors concluded that losses were not increasing
quickly enough to exceed the structural protection that is built
into senior CMBS bond issues. Incorporating investments not
represented in the index such as high-yield, floating-rate and
non-U.S.-dollar-denominated bonds would also add diversification.
In viewing out-of-index bonds as volatility reducers
as well as income generators, doubters of a global recovery
would benefit from the fact that the correlation between highyield
bonds and Treasurys has varied from—0.25 to 0.50 in
recent years (see Figure 2).
Consider also that the government-dominated “total market”
Barclays Aggregate doesn’t include TIPS—which doesn’t
really make much sense. Including those would more accurately
represent an investor’s opportunity set in fixed income.
The single greatest risk to bonds is inflation, so if you were going
to make a “total” bond index that offered investors the greatest
Figure 2
BarCap
Agg
BarCap Agg 1.00
BarCap
Govt
BarCap Govt 0.86 1.00
Bond Index Correlation Matrix, April 2004 - March 2011
BarCap
Credit
BarCap Credit 0.79 0.40 1.00
BarCap
Mortg
BarCap Mortg 0.87 0.87 0.46 1.00
BarCap
TIPS
BarCap TIPS 0.52 0.31 0.43 0.46 1.00
BarCap
HiYield
BarCap HiYield 0.02 -0.48 0.52 -0.23 0.35 1.00
CS Lev Loan -0.24 -0.67 0.25 -0.43 0.26 0.94 1.00
Citi WGBI
Unhedged
Citi WGBI
Unhedged 0.65 0.61 0.56 0.49 0.33 -0.02 -0.28 1.00
JPM GBl EM
Glb Divr
JPM GBl
EM Glb Divr 0.25 -0.08 0.58 -0.04 0.20 0.66 0.50 0.45 1.00
CS Lev
Loan
Source: eVestment Alliance
Note: CS Lev Loan is the Credit Suisse Leveraged Loan Index and JPM GBI EM Glb Divr is the JPMorgan GBI Emerging Markets Global Diversified Index.
diversification bang for their buck, a well-structured one would
certainly want to include securities that hedge that risk.
Finally, unrelated to allocation, is the issue of duration.
Recognizing indexes as strategies is even more important for
institutional investors managing defined benefit plans where
the best benchmark isn’t a “best-of-breed” bond index but
rather the liabilities themselves. Why have a benchmark like
the Barclays Aggregate, which has a duration that is intermediate
in nature, when your liabilities are probably much longer?
Liability schedules have significant positive convexity, so it
would be advisable to start with an index that also has positive
convexity. Convexity refers to the relationship between a
bond’s price and its yield; bonds demonstrating greater degrees
of convexity are more affected by changes in interest rates.
Know Thy Index
Get to know your index. In all likelihood, you own a fund
that follows the Barclays Aggregate. The fixed-income market
today is at a historic juncture, and a strong argument for
even greater diversification can be made. The risk of owning
U.S. government debt is as great as at any time since the
www.journalofindexes.com
January / February 2012
63

1950s. With a largely government portfolio, you’re taking
on the risk of the government’s credit rating as well as the
risk of rising interest rates. Whether the U.S. will default on
its debt or pseudo-default by allowing inflation to run rampant
is thought-provoking, but nobody really knows what
will actually happen. Such things are notoriously difficult
to predict, as reflected by John Kenneth Galbraith’s statement
that “There are two types of economists: those who
don’t know, and those who don’t know they don’t know.”
We can hedge that risk by taking advantage of out-ofindex
opportunities or adding slices of the broader opportunity
set of fixed-income investments to our portfolios.
Thinking prospectively about what the post-debt ceiling
condition of the fixed-income markets will look like, it
makes sense to become benchmark agnostic and look
beyond the Barclays Capital U.S. Aggregate Bond Index.
Endnotes
1. Babson Capital Perspectives, “Getting To Know Your Benchmark: The Barclays Capital U.S. Aggregate Bond Index,” February 2011.
2. Daniel Kruger and Bryan Keogh, “Obama Pays More Than Buffett as U.S. Risks AAA Rating,” Bloomberg.com, March 22, 2010. http://www.bloomberg.com/apps/news?p
id=newsarchive&sid=azz5FiyZHvMY
Laipply continued from page 37
turnover and save on transaction costs.
Alternatively, suppose the portfolio manager selected a
portfolio solution with 0.60 bps per year mean PTE and an
estimated 1.8 bps per year of transaction costs corresponding
to 44 percent annualized turnover (Portfolio B in Figure 8).
Similar to Figure 7, Figure 9 shows the backtested results of the
realized tracking error for this solution versus full replication.
Although this particular solution did incur elevated
realized tracking error during the onset of the financial crisis,
it is likely that anything other than the fully replicated
portfolio would have done so for the reasons discussed.
Ultimately, however, the chosen solution successfully balanced
transaction costs and PTE across both low- and
high-volatility environments, and it provided lower total
realized tracking error than a full replication solution over
the backtested horizon. This example illustrates the pronounced
impact that market volatility may have on bid/
offer spreads and transaction costs (even in the relatively
liquid market for U.S. Treasurys), greatly hindering a portfolio
manager’s ability to meet tracking objectives.
Similar to the preceding example, this data could be used
to develop a set of portfolio management guidelines. Using
these guidelines, the portfolio manager could position and
manage this particular Treasury fund across varying market
conditions and volatility regimes, trading off PTE, transaction
costs and historical realized tracking error.
An Alternative Approach
The empirical results of the backtested simulations suggest
an approach to fixed-income index portfolio management
that would incorporate a variety of potential volatility regimes
and their impact on common and idiosyncratic risks and
transaction costs. Such an approach would have properties
similar to that of an option-based or insurance-based strategy
designed to manage the portfolio to the tightest realized
expected tracking error across a range of market conditions.
Using this alternative approach, a portfolio manager
would examine the trade-offs among the following three
dimensions and attempt to choose an optimal portfolio
construction/management strategy:
• Projected tracking error
(based on risk model forecasts)
• Historical realized tracking error for different
portfolio solutions (based on empirical data)
• Transaction costs
(based on forecasts and empirical data)
The strategy would be monitored and adjusted accordingly
relative to market conditions.
The approach to fixed-income index portfolio management
described herein moves beyond traditional optimized
solutions that rely heavily on historical correlations and static
assumptions about idiosyncratic risk and transaction costs.
To obtain efficient index exposure, it is crucial to explicitly
recognize and quantify the impact of market volatility on the
stability of correlations as well as the level of idiosyncratic risk
and transaction costs. While stratified sampling techniques
seek to reduce dependency on correlation estimates, an
option-based approach seeks to also intelligently trade off
idiosyncratic risk and transaction costs through sample size
during lower-volatility periods so as to insulate the portfolio
against higher-volatility market environments later on.
With a more thorough understanding of the interplay
between market volatility, transaction costs, and potential
and realized tracking error, portfolio managers are better
equipped to manage index portfolios and minimize realized
tracking error across a variety of potential volatility
regimes and market conditions.
Endnotes
1. For a discussion on portfolio rebalancing under varying market conditions, see Lydia Chan and Sunder Ramkumar, “Efficient Portfolio Rebalancing
in Normal and Stressed Markets,” Investment Insights 13 no. 3 (September 2010) BlackRock Inc., New York.
2. Note that portfolios were constructed through risk-model-based optimizations. Stratified sampling techniques were not employed to simplify the
implementation of the simulations.
64 January / February 2012

ETFs
PLUS A WHOLE LOT MORE
Subscriptions: www.indexuniverse.com/subscriptions. Advertising and Reprints Inquiries: 415.659.9029

From The Think Tank
HUMOR
Not Quite Rocket Science
By Heather Bell
Solving the Greek
debt crisis
When you think about it, it’s kind of
ironic. There’s a very valid argument to be
made that Greek culture laid the roots for
the existence of today’s developed-market
democracies. That’s right: The developed
free world as we know it owes its existence
to a collection of islands with less square
mileage than New York state.
At the same time, modern Greece’s
failed economic policies could kick off the
next financial apocalypse. Greece giveth,
and Greece taketh away?
An October 2010 article in Vanity Fair by
Michael Lewis offered a revealing look at a
country where tax evasion is the norm, workers
for the (bankrupt) state-run railroad are
among the best-paid citizens, government
financial statistics are whatever the current
regime thinks they should be and a bunch of
monks can walk away with the $1 billion in
federally owned land without anyone being
able to fully explain how they did it.
Dysfunctional doesn’t even begin to
cover the situation. Dr. Phil probably falls
asleep at night dreaming about leading
an intervention with Christine LaGarde,
Angela Merkel and Nicolas Sarkozy.
(Greece: “Everything is under control. I
don’t have a problem!” LaGarde: “We can’t
continue to watch you binge like a sailor on
shore leave in Bangkok. We’re cutting you
off, because we love you—and because we
don’t want you to trigger the collapse of
Western civilization as we know it.” Greece:
“You’re not the boss of me!”)
With Greeks rioting to protest proposed
reforms, Europe’s central banks wringing
their hands and all sorts of mean people
saying we should just let the country default
on its debts or demand further austerity
measures, we at the Journal of Indexes have
put our heads together to humbly suggest a
few possible solutions:
1. Eat more yogurt. Greek yogurt
is what all the healthy people are eating
these days—or so the guy at Whole
Foods told me. If we all make a conscious
choice to eat nutritious Greek
yogurt, we’ll take a huge chunk out of
Greece’s $400 billion-plus national debt
in no time, in addition to building stronger
bones and staving off osteoporosis!
2. Sell off a few islands. There were
rumors a while back that the country would
take just such an action, but the Greek
government quickly squelched them. But
is it really such a bad idea? According to
Wikipedia, Greece is made up of more
than 6,000 islands. Six thousand! They’re
practically commodities—or would be if
Greece could export them. But what could
the Greeks possibly be doing with all of
them? Some of those islands don’t even
have people living on them. In fact, we
ordinary citizens here in the U.S. would
probably take up a collection to buy one
of the islands just so we could exile some
choice people there, like Donald Trump,
Nancy Grace and the Kardashians.
3. Ask the Bill & Melinda Gates
Foundation for a really big loan. Sure, the
foundation is currently using its dollars
to keep children from dying in the Third
World, but we’re talking about preserving
a way of life here! If the Greek dream
of retiring at 50 (if you’re a woman) or 55
(if you’re a guy) fizzles out, it’s a slippery
slope to preventing medical professionals
from taking home toilet paper from
hospitals and teachers from calling in
sick when they’d rather not work.
4. Convert all Greek sovereign debt
into IOUs. Written on the backs of cocktail
napkins. Greece promises to pay you
back. It really really does. Just as soon as
it gets paid next week. And by the way,
can it crash on your couch tonight? Just
for one night . . .
5. Send out a chain email from an
Athenian prince, requesting recipients’
bank account information. What? You
mean it didn’t work for that guy in Nigeria?
66
January / February 2012

VIG & VYM
Vanguard Dividend Appreciation ETF
Vanguard High Dividend Yield ETF
Are you Vanguarding ® your clients’ portfolios?
If your clients are looking for a fund that has delivered sustained dividend growth, look no
further than VIG. VYM is for those who are seeking high dividend yields. And both offer
expense ratios that are 61% lower than the industry average.* No matter what their
financial goals are, Vanguard can help your clients reach them.
Take a closer look at advisors.vanguard.com/ ETF
800-523-8845
All investments are subject to risk. Vanguard funds are not insured or guaranteed.
To buy or sell Vanguard ETFs, contact your financial advisor. Usual commissions apply. Not redeemable. Market price may be
more or less than NAV.
For more information about Vanguard ETF Shares, visit advisors.vanguard.com/ETF, call 800-523-8845, or contact
your broker to obtain a prospectus. Investment objectives, risks, charges, expenses, and other important information
are contained in the prospectus; read and consider it carefully before investing.
*Source: Morningstar as of 08/31/2011. Based on 2011 industry average expense ratio of 0.46% for equity income ETFs and
Vanguard VIG and VYM expense ratios of 0.18%.
© 2012 The Vanguard Group, Inc. All rights reserved. U.S. Pat. No. 6,879,964 B2; 7,337,138; 7,720,749; 7,925,573. Vanguard Marketing Corporation, Distributor.
Follow us @Vanguard_FA for important insights, news and education.

Is your focus in the right place?
Risk or fortune? Whatever lies ahead,
we’re here to help you see what matters most.
Russell Indexes. Opportunity Re-Defined.
+1 877 503 6437
www.russell.com/indexes
Russell Investments is a Washington, USA Corporation which operates through subsidiaries worldwide and is a subsidiary of The Northwestern Mutual Life Insurance Company.
Russell Investments is the owner of the trademarks, service marks and copyrights related to its respective indexes. Indexes are unmanaged and cannot be invested in directly.
Copyright © Russell Investments 2011. All rights reserved.