This commentary is written by Martin Fridson, a high-yield market veteran who is chief investment officer of Lehmann Livian Fridson Advisors LLC as well as a contributing analyst to S&P Global Market Intelligence.
Until now, high-yield portfolio managers who would like to complement bottom-up security selection with strategic sector allocation have had few tools at their disposal.
The ICE BofA US High Yield Index's manager does provide a Composite Rating for each industry. Ratings are somewhat helpful in managing default risk, which consists of both default probability and expected loss given default. A portfolio manager's selection of specific issues and their weightings within an industry, however, may not have the same rating mix as the corresponding industry subindex.
Another complication is that two industries with identical Composite Ratings may not have equivalent default risks. This simplified example explains why: Subindex A consists of one bond rated BB/Ba2/BB and one rated CCC/Caa2/CCC. In the ICE Indices, LLC system, the subindex's Composite Rating would be B2. Assuming all the ratings involved in this example correctly reflect default risk, Subindex A has more default risk than Subindex B, which consists of two bonds, both rated B/B2/B. The reason is that default risk does not increase on a straight-line basis as ratings decline but rather at an accelerating rate. Therefore, the average default risk of a BB2 and a CCC2 bond exceeds that of the average of two B2 bonds (see note 1).
A more problematic limitation of Composite Ratings as a portfolio management tool is that they address only default risk. A high-yield manager whose portfolio is marked to market and who is evaluated on the basis of risk-adjusted return is not solely concerned with default losses. As we shall show, a reallocation from one industry to another with the same Composite Rating may entail a substantial increase or decrease in expected volatility, which is the risk in the risk-adjusted return calculation.
MORE FRIDSON: Sub-4% does not mark the end of 'high'-yield
Portfolio managers also have available the weighted-average option-adjusted spread for each subindustry. This measure provides some information about the reward for recurring a given industry's risk. In addition, it is popularly regarded as a "market rating" that some investors may prefer to agency ratings. Like agency ratings, however, the OAS gives no clear indication of an industry's expected volatility. See below for evidence on this point. Furthermore, neither measure addresses yet another aspect of risk management, the ability to reduce portfolio-wide volatility through diversification.
Exploiting industry volatility measurement
To address the existing information gap in comprehensive high-yield risk management, we calculated standard deviations of monthly total returns over the period 2006-2020 for the 20 largest (by face amount) industry subindexes of the ICE BofA US High Yield Index. Choosing the historical observation period involves a trade-off. Returns are available for most industry subindexes as early as 1997. In some cases, though, data nearly a quarter-century old may not be indicative of current industry risk.
For example, the high-yield Telecommunications sector soared during the monetary regime that also spurred the "dot.com bubble" of the late 1990s. The boom in bonds of early-stage telecom ventures turned into a crash during the Tech Wreck in the early years of the new millennium. Massive defaults contributed to a 44% decline in the ICE BofA US High Yield Telecommunications Index's issue count between November 2000 and April 2002, a period in which the number of bonds in the US High Yield Excluding Telecommunications Index grew by 23%. In our judgment, the optimal historical observation period for measuring industry volatility excludes the earliest available data but includes data for the Great Recession that began in January 2008 and the two years leading up to it, as well as the recession year 2020.
The table below ranks the 20 industries in our test sample according to monthly standard deviation of total return over the period 2006-2020. A key observation is that the ranking by this volatility measure provides new information about industry risk. This is shown by the highly imperfect correlation between the rankings by volatility and by rating, whether of the agency or the market variety.
For example, industries with Composite Ratings in the BB tier are distributed as evenly as they possibly could be for an odd-numbered total of nine. The top (riskiest) half of the standard deviation ranking contains five of those industries, while four are found in the bottom (least risky) half. Similarly, five of the industries with Composite Ratings in the B tier are in the top half of the volatility rankings and six in the bottom half. (Note: Super Retailing does not contain supermarket bonds, as its name might suggest, but rather issues of department store chains, discount retailers, and specialty store operators.)
The connection between volatility-based risk and ratings is somewhat stronger when we turn to market ratings, i.e., OAS. The correlation between those two columns in the table is a not inconsequential 32.2%. The two widest (highest-risk) spreads are found in the top (riskiest) half of the standard deviation ranking, while the two narrowest (least risky) spreads are found in the bottom (least-risky) half of the standard deviation ranking. Even so, there is far from a perfect correspondence between volatility and spread. Most strikingly, the riskiest industry by standard deviation, Automotive & Auto Parts, has a narrower spread (+327 bps) than the least risky (Containers), at +329 bps.
On the whole, the statistics presented in the table above uphold our claim of introducing genuinely new information to high-yield portfolio managers seeking to maximize risk-adjusted return. Armed with this new information, PMs are no longer limited to managing risk solely on the basis of default probability and expected loss given default. Now they can also take industry behavior into account and manage risk on a dimension that is probably more important to them in most measurement periods, namely, the denominator of the return/standard deviation ratio.
The final column in the preceding table further underscores this new information's value in the risk management process. It displays a metric we report in our monthly recaps of high-yield performance, to wit, how each industry is priced on a rating-for-rating basis versus its peers. (For background, see "New industry analysis shows defensives too tight.") The numbers indicate that, for example, the market considers a Metals & Mining bond 12.34% riskier than the average bond of its rating. Conversely, the market, on average, considers a Consumer Products bond only 89.25% as risky as the average like-rated bond, as indicated by its -10.75% score (100.00% - 10.75% = 89.25%).
There is a comparatively tight fit between this metric and industry standard deviation of return. The correlation between the two series is a strong 57.2%. Furthermore, eight of the 10 least-risky industries, measured by standard deviation, have negative scores, meaning that they are priced tight to their peer group. Similarly, of the 10 riskiest industries based on volatility, eight have positive scores. These findings tell us that the market incorporates differences in total return volatility into industry spreads, even though portfolio managers may not describe their valuation practices in those terms. PMs may simply divide the industry universe into "cyclical" and "defensive" categories, based on informal observations made during recessionary periods.
Optimizing industry diversification
Industry volatility numbers represent a new item in the portfolio manager's toolkit, but they provide only part of the benefit that can be extracted from industry return data. PMs can decrease risk by utilizing the closest thing to a free lunch that the securities market offers. That is, they can reduce portfolio-wide volatility by combining assets that have low correlations with one another.
To facilitate the implementation of such a strategy, we calculated the 190 industry-to-industry monthly total return correlations of the high-yield market's 20 largest industries for the period 2006-2020. The table below presents an analysis aimed at translating the mass of data into practical steps to capture the potential for industry diversification.
The left-hand portion of the table ranks the 20 industries from best to worst diversifying asset, based on the number of other industries with which an industry has a total return correlation of 80.0% or higher. By this measure, the best industry to emphasize in order to promote diversification is Automotive & Auto Parts, which does not correlate as highly as 80.0% with any other major industry. At the opposite end of the spectrum, Technology, Super Retailing, and Services are the least helpful industries to own from the standpoint of obtaining diversification benefits. Each of those three has a correlation of 80.0% or more with 10 other industries.
In the middle portion of the table, industries are again ranked from best to worst diversifiers, but in this instance by the number of industries with which they have total return correlations of 70.0% or less. By this measure, Automotive & Auto Parts again provides the greatest diversification benefit. It has a 70.0% or less correlation with 18 of the other 19 industries. Two of the three industries that ranked at the bottom by the previous criterion — Technology and Super Retailing — tied with Homebuilders & Real Estate according to this second ranking method.
The right-hand portion of the table presents summary data. We rank the industries by the average of their ranks in the other two lists. For example, Healthcare's No. 14 and No. 7 ranks in the left-hand and middle portions of the table produce an average of 10.5, placing it 11th in the summary ranking that portfolio managers should use as their guide to decide which industries to emphasize or deemphasize. Note that the No. 19 final ranking of Technology and Super Retailing is not the average of their No. 18 and No. 18 scores in the other two rankings. Rather, their average scores of 18 put them in last place (a tie for No. 19) in the final ranking.
Minimize industry volatility or maximize diversification?
Portfolio managers who examine this study's two tables together will quickly notice a potentially dismaying pattern. The industries that are the most attractive to own in terms of enhancing diversification (ranking high on the second table) are generally the least attractive to own from the standpoint of volatility (ranking high on the first table). Most strikingly, the Automotive & Auto Parts industry is both the best industry to own for diversification terms and the worst to own if the goal is to hold industries with stable returns.
Some investment organizations may not be put off by high correlations among comparatively stable industries. If so, they will opt to emphasize industries such as Food, Beverage & Tobacco and Containers, which combine low standard deviations of returns with minimal diversification benefits. The total return correlation between those two particular industries is one of the highest in the analysis, at 90.7%.
An alternative strategy is to emphasize the industries that offer the best trade-off between low volatility and high diversification benefit. Two industries stand out in this regard. Cable & Satellite TV ranks among both the 20% least volatile industries and the 20% best diversifiers. To be sure, the market extracts a price for those favorable characteristics. Cable & Satellite TV bonds are currently priced on average at only 87.81% of the spread on like-rated bonds from all industries (100.0% -12.19% = 87.81%). The No.1 pick for emphasis according to the volatility-diversification trade-off is the Utility industry. It ranks just below the least volatile quartile and in the top 20% for diversification benefits. Emphasizing this industry requires only a minor sacrifice in risk premium. Option-adjusted spreads on Utility bonds are on average 99.08% as wide as on spreads of like-rated bonds from all industries (100.00% - 0.92% = 99.08%).
Research assistance by Bach Ho and Ducheng Peng.
ICE BofA Index System data is used by permission. Copyright © 2021 ICE Data Services. The use of the above in no way implies that ICE Data Services or any of its affiliates endorses the views or interpretation or the use of such information or acts as any endorsement of Lehmann Livian Fridson Advisors LLC's use of such information. The information is provided "as is" and none of ICE Data Services or any of its affiliates warrants the accuracy or completeness of the information.
1. Readers can rest assured that the actual industry subindexes discussed in this piece are not subject to the huge statistical noise that would be the case for the hypothetical two-bond subindex in our illustration of non-equivalent default risk for like-rated subindexes. The analysis in this study focuses on the ICE BofA US High Yield Index's 20 largest industries by face amount. We excluded small industries specifically to avoid the problem of statistical noise. Collectively, these industries account for 87% of the index's total face amount. The smallest industry by number of issues (Consumer Products) consists of 32 bonds as of February 2021. That number satisfies the common minimum standard of 30 items to constitute a scientific sample. Energy, with 328 constituents, far exceeds any other industry's issue count, but in all, 16 of our 20 test industries contain more than 50 bonds.