• Allocating by debt outstanding creates a lopsided yield distribution, with a barbell-like overconcentration both in relatively safe-haven assets with a limited income profile and in fundamentally risky debt profiles.
• The benchmark index does not foster diversification, with a two-thirds weighting to government affiliated bonds and high correlation between the two largest sectors.
• Investors should consider a multi-sector bond strategy filtered for opportunity rather than indebtedness. Such a strategy could address the investment universe screened by yield, quality and liquidity.

The growth of the US bond market

The United States, initially as the Continental Congress, first incurred debt in 1776 when it borrowed funds to finance the Revolutionary War.¹ Total Treasury debt remained fairly small in the first half of the 19th century but rose sharply with the Civil War and again with World War I (Figure 1). After declining slightly, the debt increased nearly threefold during the Great Depression and exploded in the 1940s as the government financed expenditures related to World War II.

From its postwar low in 1949, outstanding Treasury debt grew gradually for nearly two decades before accelerating at the time of the Vietnam War and during the subsequent period of high inflation. In the 1980s, the growth of the stock of debt picked up further, spurred by the tax cuts and rapid increases in defence spending of the decade.² America's continuing and growing budget deficits, combined with the unprecedented government intervention in US financial markets in 2008 - including the mortgage-backed securities (MBS) issued by Fannie Mae and Freddie Mac, have driven the explosion in US government debt outstanding.

FIGURE 1: US GOVERNMENT DEBT OUTSTANDING
1790-2017

Source: US Department of the Treasury. Data as at 31 August 2017.

The birth of a bond benchmark

On 26 May 1896, Charles Dow created what we now know as the Dow Jones Industrial Average. But total return bond indices weren't developed until 1973. The growth of asset allocation in portfolio management in the 1970s necessitated a measure of bond performance, hence the need for a bond return benchmark. Until then, bonds were rarely traded, and most investors just bought and held them to maturity.

In 1973 Art Lipson and colleague John Roundtree at Kuhn, Loeb & Co. created what later would be called the Agg. Lehman Brothers purchased Kuhn, Loeb & Co. at the end of 1977, and Barclays Capital took over the index business of the now-defunct Lehman Brothers in 2008. In 2016 Bloomberg acquired Barclays Risk Analytics and Index Solutions Ltd., giving us what is currently called the Bloomberg Barclays US Aggregate Bond Index - the Agg.

While manually calculating the value of a 30-stock index (the Dow Jones Industrial Average) was feasible, it was the appearance of computers in the financial industry in the 1970s that enabled Mr Lipson, an engineering major in college, to develop a program to keep track of more than 3,500 bonds. These bonds had a total value of $221 billion at inception.³ Now, the Agg comprises a total of more than 9,300 bonds and is worth nearly $20 trillion (Figure 2).

FIGURE 2: BLOOMBERG BARCLAYS US AGGREGATE BOND INDEX MARKET VALUE
1976-2017

Bloomberg, data as at 31 August 2017.

The Agg is more than an index. It is the basis for financial products that represent a majority of fixed income allocations for many investors. For example, the iShares Core US Aggregate Bond ETF, launched 26 September 2003, is the world's largest bond ETF,⁴ with assets of $45 billion. That's equal to about 10% of the total assets invested in US-listed domestic fixed income ETFs - 249 funds in all.

Challenges with the bond benchmark: sector diversification

The financial crisis of 2008 and subsequent government intervention has changed the complexion of the US bond market. In 2007 the Agg was 22% US treasuries, but that has increased to 37% today. Factoring in debt issued by government agencies⁵ and mortgage-backed securities, the total government exposure is now over 70% (Figure 3). Additionally, the Agg weightings historically exhibit high correlations among the top components (Figure 4). The correlation of the top two components, US treasuries and MBS, is 81% with minimal exposure to those components with low-cross-correlations.

By default, bond market investors who use the Agg have their largest position in the lower left of the fixed-income risk-reward profile (Figure 5): these types of bonds have historically exhibited low return and low volatility.

FIGURE 3: BLOOMBERG BARCLAYS US AGGREGATE BOND INDEX, SECTOR BREAKDOWN (%)

Bloomberg, data as at 31 August 2017.

FIGURE 4: AGG COMPONENTS, CORRELATION
(31 JANUARY 2006 TO 31 AUGUST 2017)

Source: Bloomberg, correlation calculated is based on monthly returns, data as at 31 August 2017. Correlation ranges from +1 to -1. Positive correlation indicates returns moving in the same direction, negative correlation indicates returns moving in opposite directions, and a correlation of 0 would indicate no relationship between the movement of the two returns. For index definitions, please refer to the end of the article.

FIGURE 5: AGG COMPONENTS, CORRELATION
2006-2017

Source: Bloomberg, data as at 31 August 2017.

The top two components of the Agg have an 81% correlation.

Opportunity: Targeting less correlated sectors

There is diversification potential if you examine other sectors of the bond market. Moving out along that risk-reward profile finds opportunities that are both less correlated (Figure 6) and have historically offered relatively higher returns (Figure 7). Sectors like US Corporate High Yield, Global Treasuries and the Emerging Market USD Aggregate historically have had much lower cross-correlations and are not found in the Agg (see Figure 6).

FIGURE 6: MULTI-SECTOR, CORRELATION
(31 JANURARY 2006 TO 31 AUGUST 2017)

Source: Bloomberg, correlation calculated is based on monthly returns, data as at 31 August 2017. Throughout this paper, Emerging Market USD Aggregate or Emerging Market Debt represents Emerging Market (EM) Sovereign debt denominated in US Dollars; Global Treasuries or Global Tsy Ex US represent non US Treasuries. For index definitions, please refer to the end of the article.

FIGURE 7: MULTI-SECTOR, RISK/RETURN PROFILE
(2006-2017)

Bloomberg, data as at 31 August 2017. For index definitions, please refer to the end of the article.

Challenges with the bond benchmark: Yield, quality and liquidity

In equity markets, investors typically rely on market-capitalisation-weighted benchmark indices. Arguably, size can be associated with quality. However, many investors have traditionally used products that track the debt-weighted Bloomberg Barclays US Aggregate Bond Index as their core fixed-income allocation, and when it comes to fixed income the largest issuers do not follow the same logic. These challenges are illustrated in a number of ways.

First, allocation by debt outstanding creates a lopsided yield distribution, with a barbell-like overconcentration between relatively safe-haven assets (eg, US treasury, Japan government bonds, highly rated/large-cap corporates), with a limited income profile on one end and fundamentally risky debt profiles (eg, Italy, Venezuela, highly indebted/ low-rated corporates) on the other end.

Secondly, archaic segmentation creates distortion in credit quality, which may present a challenge for investors, especially within the investment grade (IG) and high yield (HY) markets. The IG market includes high-rated/low-yielding bonds, while the HY market includes issuers with bloated capital structures and spotty liquidity. While the IG market may have attractive characteristics for investors, there is a large range of issuers in the market, with only about 10% of bonds actually trading at the average yield. For example, certain highly rated, liquid bonds issued by large companies trade at essentially government bond yields.

Similarly, the developed market (DM) and emerging market (EM) distinctions create anomalies. Highly indebted Italy - with a 133% debt/gross domestic product (GDP) ratio - and Portugal (130%) are classified as developed markets and Chile (21%) and South Korea (39%) are classified as emerging markets.⁶

Lastly, post-crisis regulation has created a broad spectrum of liquidity. High-quality securities - US Treasuries, Agency mortgage-backed securities - are generally more liquid and offer lower yields, while riskier, higher yielding securities may trade less frequently, if at all. Trading in these less liquid markets can create high volatility.

Opportunity: cleaning up the bond benchmark

We believe a multi-sector bond strategy filtered for opportunity rather than indebtedness may provide a better balance of yield, quality and liquidity than the benchmark. Such a strategy could address the
investment universe screened by:

• Yield: Include multiple sectors throughout the US and around the globe, including some that are not part of the Agg.
• Quality: Avoid the "tails of the market" by removing certain sectors (eg, Japan, highest grade/low-yield corporates) that offer no risk premium and low-quality tiers (eg, Venezuela, lowest grade/high-yield corporates) that could potentially have massive downside risk.
• Liquidity: Focus on issues with sufficient tradability to provide investors with liquidity when they need it, with volatility that is tolerable.

Additionally, by incorporating the points above, investors seeking higher returns could be wellserved to move out along that risk/return profile, appropriately reweighting components and, importantly, filtering those components to reduce market factor idiosyncrasies, as seen in Figure 8.

FIGURE 8: A MULTI-SECTOR BOND STRATEGY
2006-2017

Here are some examples of how yield, quality and liquidity filters could be implemented at a sector level, keeping in mind that these strategies may not be appropriate for all investors:

Bond credit ratings are assigned by third-party agencies and are divided into categories ranging from AAA (highest) to D (lowest). Credit ratings are subjective opinions of the credit rating agency and not statements of fact, may become stale and are subject to change.

Conclusion

Investors have traditionally used products thattrack the Bloomberg Barclays US Aggregate Bond Index as their core fixed-income allocation. As we enter a new rate regime, investors may need to adjust their fixed-income portfolios to avoid overconcentration and minimise interest rate risk. Allocation by debt outstanding creates a lopsided yield distribution, with a barbelllike overconcentration both in relatively safehaven assets with a limited income profile and
in fundamentally risky debt profiles. The bond benchmark does not foster diversification with historically high correlations among the largest components. Additionally, archaic segmentation creates distortion in credit quality. Investors should consider a multi-sector bond strategy filtered for opportunity rather than indebtedness.
Such a strategy could address the investment universe screened by yield, quality and liquidity. In the current lower rate environment investors seeking higher returns could be well-served by appropriately reweighting components and, importantly, filtering those components to reduce market factor idiosyncrasies.

¹ Rafael A. Bayley, The National Loans of the United States of America from July 4, 1776 to June 30, 1880, as Prepared for the Tenth Census of the United States (Washington, DC: U.S. Government Printing Office, 1883).
² The Treasury Securities Market: Overview and Recent Developments http://eh.net/database/u-s-government-bond-trading-database-1776-1835/.
³ "Barclays Agg Had Modest Origin," The Wall Street Journal, April 2, 2013.
⁴ The iShares ETFs are not sponsored, endorsed, issued, sold or promoted by Columbia Threadneedle Investments.
⁵ Government agencies includes Agencies, Supranational and Local Authorities as shown in Figure 3.
⁶ International Monetary Fund (IMF), data as of April 2017.

Download PDF