• U.S. Credit Cycle: It is becoming increasingly clear that we are now in the latter stages of the multi-year credit cycle. Credit investors should lower their return expectations going forward.
  • Bond Benchmarks: The traditional bond indices that bond managers use to measure their investment performance are flawed for a simple reason - they use market-capitalization based weightings that give more influence to riskier issuers.
  • Alternative Benchmarks: Using an asset allocation scheme that weights corporate bond sectors by a measure of credit risk (duration-times-spread) produces portfolios that are likely to outperform traditional corporate bond indices in the next credit bear market.

Chart 1
A Bearish Turn In The Credit Cycle
Chart 1

Fullscreen        Interactive Chart

Feature

Prior to this year, the post-crisis years were tremendous for U.S. corporate credit markets. Investment Grade (IG) corporate bonds returned an annualized 6.5% over the five-year period between 2010 and 2014, while High-Yield (HY) corporates returned 9.0%, both solidly outperforming U.S. Treasuries. The story has been quite different so far in 2015. IG has posted a negative year-to-date total return of -0.84% and HY total returns are essentially zero, both substantially underperforming Treasuries.

It is no surprise that corporate credit performance began to deteriorate after our own measures of corporate balance sheet health and monetary conditions began to turn in a more bearish direction for credit in the latter half of 2014 (Chart 1). Domestic monetary conditions have tightened significantly over the past year, led by the end of the Federal Reserve's Quantitative Easing (QE) program and subsequent strengthening of the U.S. dollar. More importantly, corporate balance sheet health has steadily deteriorated on the back of rising leverage and weaker profit growth.

It is clear that the U.S. is now in the latter stages of the multi-year credit cycle - the period when rising corporate leverage negatively impacts returns to corporate debt as investors demand higher risk premia (wider credit spreads) to compensate for the greater volatility created by higher leverage.

This week, we are publishing the first in a series of occasional Special Reports that will discuss portfolio strategies that we believe bond investors can use to navigate a more challenging backdrop for credit returns. In this initial report, we will discuss how using an alternative method to determine asset allocation among IG corporate sectors - weighting exposures by underlying spread risk and not by market capitalization as is used in traditional bond benchmark indices - can provide a way to better diversify risk exposures in credit portfolios and boost risk-adjusted returns.

The Problem With Using Traditional Bond Indices To Manage Portfolio Risk

Bond investors usually make asset allocation decisions in their portfolios with respect to an index that is designed to proxy the passive return of owning a specific fixed income asset class, like corporates or Treasuries. This provides a performance benchmark that a bond manager can be measured against, under the assumption that the total return of an index is a fair estimate of the total return of the entire investible bond market and, thus, the full opportunity set available to the bond manager.1

The most widely-used method for weighting individual bonds within bond indices has been based on market capitalization, where the current market value of the bond determines its size in the index, similar to the basic benchmark indices used in equity markets. However, this creates an undesirable situation where issuers that take on increasing amounts of debt get a larger weight within bond indices, even though those are the issuers that are often more at risk of future credit strains and worsening credit ratings as they service that larger stock of debt. In other words, the riskier credits get the larger weights - what has been colorfully referred to as "the bums problem" of bond indices.2

The logical conclusion is that while a market-capitalization-based index does represent a fair assessment of the investible size of the underlying fixed income asset class, it does not always produce an accurate reflection of the risk of the same asset class.

Global government bond investors have been well aware of this problem, where a country like Japan with a massive stock of debt ends up with a very large weight in global government bond indices, despite poor credit metrics. This has led investors to ask bond index providers to produce alternative index weighting schemes that generate smaller exposures to the heaviest borrowers, such as weighting countries in an index by GDP.

Other investors have taken the more extreme decision to become "benchmark-agnostic" by eschewing the use of traditional bond indices altogether when making asset allocation decisions. In theory, this approach gives a bond manager the maximum flexibility to manage duration and credit risk in a bond portfolio with little, if any, adherence to a bond index. In practice, however, these strategies often use a high degree of portfolio leverage via derivatives to reduce/eliminate "unwanted" risks, while at the same time putting the onus on the bond manager to accurately measure portfolio risk, which involves estimating unstable correlations and volatilities. The mixed performance of so-called "unconstrained" bond funds over the past few years suggests that this is much easier said than done.3

While government bond managers now have several alternative bond indices to use in managing their portfolios, corporate bond managers have few options when it comes to alternative index weighting schemes that can help mitigate the risks inherent in the conventional bond indices.

As can be seen in the second panel of Chart 2, there has been a steady increase in the average modified adjusted duration (the sensitivity of bond prices to changes in bond yields) and average spread duration (the sensitivity of bond prices to changes in spreads) of the Barclays U.S. IG Corporate index over the past fifteen years. This is a result of the steady decline in government bond yields over the same period, particularly in the post-crisis years where the Fed was driving down Treasury yields via its QE programs. This is also a glaring example of the problem of using market-capitalization bond indices as investment performance benchmarks.

Chart 3
A Lot Of Duration Risk In Corporates
Chart 3

Fullscreen        Interactive Chart

While the indices may be indicative of the full investible universe of bonds available to the bond manager, the indices may not adequately reflect the risk tolerances of the end-investor whose funds are being managed. An allocation to IG corporates now has a very high sensitivity to changes in interest rates for an asset class that traditionally offers a limited spread cushion to absorb losses from yield increases.

Of course, corporate bond managers can (and usually do) mitigate that exposure to changes in interest rates via some degree of duration management at the full portfolio level. However, that comes at a cost for managers who must adhere to yield "bogeys" or risk constraints when constructing portfolios.

A bond manager that wants to reduce the interest rate duration of his/her portfolio has essentially three choices: (i) reduce exposure to corporates and put the proceeds into cash, (ii) sell longer-maturity bonds and buy shorter-maturity bonds, or (iii) use derivatives like Treasury futures or interest rate swaps to reduce overall portfolio duration as an overlay position. The first two options come at a cost in terms of a reduced portfolio yield, while the use of the final option may be limited by portfolio guidelines concerning the usage of derivatives or overall portfolio leverage. There is no free lunch to protect a credit portfolio from rising interest rates.

Similar limitations apply to managing credit risk in a corporate portfolio. Corporate bonds are a non-normally distributed asset class with negatively-skewed returns and fat tails. This is not news to any corporate bond manager, who understands that when the credit cycle turns negative, there are usually few places to hide within the corporate bond universe - all spreads tend to widen together in bear markets. But, again, efforts to protect a portfolio from the effects of a credit market downturn come with a cost in terms of adherence to risk limits.

For example, a bond manager whose performance was benchmarked to a market-capitalization weighted bond index could choose to underweight individual issuers, or even entire sectors, where credit quality was deemed to be worst. However, those issuers often have the largest weights in the benchmark because of the aforementioned "bums problem" with conventional bond indices. Underweighting the larger issuers can take-up a significant portion of the risk budget (tracking error versus the benchmark) for a bond manager ... especially if that manager is forced to use a good chunk of the risk budget to reduce the interest rate duration of his/her benchmarked bond portfolio.

Bottom Line: The traditional bond indices that bond managers use to measure their investment performance are flawed for a simple reason - they use market-capitalization based weightings that give more influence to riskier issuers.

Building A Better Mousetrap: Using Duration-Times-Spread To Generate Portfolio Allocations

So what is a bond manager to do when there are few alternative bond indices available to the market-capitalization indices, as is the case with corporates? We suggest using a risk-weighted framework that weights individual bond exposures by what we believe to be the best measure of credit risk: duration-times-spread, or DTS.4

DTS is simply the product of multiplying the credit spread of a bond times the spread duration of the bond. The underlying logic of using DTS as a measure of credit risk is that credits that trade at wider spreads are inherently riskier since they tend to experience greater spread changes, often because of sector-specific risks. Two bonds that have the same spread duration, but that trade at much different spread levels, usually exhibit much different excess return volatilities. By scaling spread duration by the actual levels of spreads, as in the DTS calculation, the return volatilities between bonds with comparable DTS levels will be much more in line.

As can be seen in Chart 3, the DTS of the Barclays IG corporate bond index at the beginning of each month has been highly correlated with the realized excess return volatility of the index during the subsequent month.5 This relationship also appears to hold at the individual sector level, as seen in Chart 4, which plots the average DTS level versus the average realized excess return volatility for each of the 47 credit sectors of the Barclays IG corporate index over the same period shown in Chart 3 (1999-2015).

Chart 3
Higher DTS = Higher Return Volatility...
Chart 3

Fullscreen        Interactive Chart

Chart 4
...Even At The Sector Level
Chart 4

Fullscreen

Given the stable relationship between DTS and risk (return volatility), we think there is value in using DTS as an alternative weighting scheme to market-capitalization weights in corporate bond portfolios. Essentially, the weight of any individual security in a portfolio is determined by that security's DTS as a share of the total sum of the DTS of all the bonds in a portfolio. Thus, each bond is weighted by its own risk, measured in a common risk framework with all other bonds in the portfolio.

We calculated this DTS-weighted portfolio for the individual credit sectors of the Barclays IG Corporate index - the results are shown in Table 1. In the table, we show the portfolio weights derived using an "aggressive" version of a DTS weighting scheme that gives the higher weight to the riskier sectors, and a "defensive" version (1/DTS) that gives lower weight to riskier sectors. We also show the actual Barclays IG index weights for comparison, along with an even simpler alternative weighting scheme that gives equal weight to each sector. Importantly, there are no volatilities or correlations that need to be estimated when constructing a portfolio this way, as we simply let the DTS determine the portfolio weights.

Table 1
Various Weighting Schemes For Investment Grade Corporate Sectors
Table 1

Fullscreen

The first main observation is an obvious one - both the aggressive and defensive DTS weighting schemes distribute the weights among the sectors much more evenly than the market-capitalization weighting scheme. This is most evident in the weighting of the Banking sector, which represents 22% of the market-capitalization weighted portfolio but only 1% and 3% of the aggressive and defensive DTS portfolios, respectively. But it is also visible in the differing weights for several of the other major sectors, most notably Capital Goods and Consumer Cyclicals.6

The second observation is that using the alternative 1/DTS weighting scheme produces weights that ensure that the product of the weight times the sector DTS is the same for each sector. In other words, each sector's contribution to the total risk of the portfolio, as measured by DTS, is equal. In theory, this should produce investment returns that are superior, in a risk-adjusted sense, to the market-capitalization weighted bond index "portfolio".

A look at Table 2 shows that this is indeed the case. Here, we show the annualized return and volatility data for the four portfolios in our study. The DTS, 1/DTS and equal-weighted portfolios all outperformed the market-capitalization index "portfolio" for the entire sample period of our study (1999 to July 2015), both in terms of absolute and excess returns and with higher Sharpe Ratios7 than the index.

Table 2
Portfolio Return & Risk Statistics For The Various Sector Weighting Schemes
Table 2

Fullscreen

We also looked at the relative performance of these portfolios during bull and bear markets for credit defined by periods when the Barclays index option-adjusted spread (OAS) was in a widening (1999-2002, 2007-2008) or narrowing (2002-2007, 2008 to 2015) phase. The results are as expected, with the defensive 1/DTS portfolio being the best performer, by any measure, in the two "down" cycles for credit markets, while generally underperforming in the two "up" cycles. Perhaps most importantly, the defensive 1/DTS portfolio produced the highest Sharpe Ratio in all four periods, largely because of the lower return volatility generated by placing lower weights on the riskiest sectors.

Bottom Line: Using an asset allocation scheme that weights corporate bond sectors by a measure of credit risk (duration-times-spread) produces portfolios that are likely to outperform traditional corporate bond indices in the next credit bear market.

Investment Conclusions & Future Research

We are encouraged by these initial results, which show how using a simple risk-weighted asset allocation scheme among corporate bond sectors - involving no estimation of correlations or volatilities - can generate portfolios that have consistently delivered superior risk-adjusted returns compared with traditional market-capitalization bond indices for IG corporates. Given our belief that we're now in the latter stages of the multi-year credit cycle in the U.S., there is considerable value in using the defensive version of our DTS-weighted portfolio as the relevant asset allocation benchmark, instead of the conventional market-capitalization weighted bond index.

Of course, there are practical considerations for a bond manager deciding whether or not implement such an alternative weighting scheme. For example, using the DTS-based weighting schemes generates sector allocations that in some cases are much larger than the index, thus there are potential liquidity issues for managers who may have difficulty finding enough bonds to match the optimal risk weights. Also, bond managers may have difficulty incorporating these alternative weighting schemes into their own risk management systems while still adhering to their own risk management guidelines. Importantly, the more defensive version of the DTS portfolio has historically generated lower volatility than the conventional corporate index, thus using this framework will likely reduce the realized tracking error of a corporate portfolio, all else equal.

In future Special Reports, we will look to extend this analysis to our broader U.S. fixed income universe by using DTS as a weighting mechanism among other spread sectors like High-Yield corporates or Asset-Backed Securities.



Robert Robis, Managing Editor
rrobis@bcaresearch.com





  • 1 For an excellent review of corporate bond indices, please see http://faculty-research.edhec.com/servlet/com.univ.collaboratif.utils.LectureFichiergw?ID_FICHIER=1328885974026
  • 2 Siegel, Laurence B. "Benchmarks and Investment Management". The Research Foundation of the Association for Investment Management and Research, August 2003.
  • 3 http://fortune.com/2015/03/25/unconstrained-bond-funds-risk/
  • 4 DTS (Duration Times Spread)" Ben Dor, Dynkin, Hyman, Houweling, van Leeuwen, and Penninga, Journal of Portfolio Management, Winter 2007.
  • 5 In our DTS calculations at the individual sector level, we use interest rate duration rather than spread duration due to data availability, as there is a longer history of interest rate duration data available for each sector in the corporate bond index. As can be seen in the bottom panel of Chart 2, there is essentially no difference in calculating DTS using interest rate duration instead of spread duration for the overall Barclays IG Corporates index.
  • 6 We've prepared several charts showing the different sector weights produced by the DTS-weighted schemes versus the market capitalization weights, which can be found in the Appendix on Pages 11-12.
  • 7 The Sharpe Ratio is defined as excess return divided by excess return volatility.



    Appendix


    Chart 5
    Lower DTS Sectors
    Chart 5

    Fullscreen        Interactive Chart

    Fullscreen

    Chart 7
    Moderate DTS Sectors
    Chart 7

    Fullscreen        Interactive Chart

    Chart 8
    Energy Related Sectors
    Chart 8

    Fullscreen        Interactive Chart