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.
How do three recently launched high-yield Environmental, Social, and Governance (ESG) indexes diverge from the standard high-yield benchmark in ratings mix, term risk, spread, and Energy concentration? We will examine these divergences, and take them into consideration as we conduct a relative return analysis, looking at ESG versus the overall high yield sector.
Background
On June 17, 2020, in response to growing demand for fixed income benchmarks that consider ESG criteria, ICE Data Indices, LLC introduced several new series of corporate and government bond indexes. As detailed below, all three U.S. high-yield USG indexes exclude bonds of companies with Sustainalytics scores of 20 or more for “Controversial Weapons Most Significant Involvement.” One of the three also excludes bonds of companies with scores of 30 or more for ESG Risk. ICE Data Indices, LLC retroactively provided daily descriptive statistics, including returns, from Dec. 31, 2016, forward.
The three U.S. high-yield ESG indexes are subsets of the ICE BofA US High Yield Index. In brief, they are constructed as follows:
- ICE US High Yield ESG Tilt Index
This index excludes bonds of companies with significant involvement in controversial weapons, and tilts the weights of remaining issues in favor of those with better (lower) ESG Risk Scores.
- ICE US High Yield Duration-Matched ESG Tilt Index
This index follows the methodology of the ESG Tilt Index (above) but further adjusts the issue weights to match as closely as possible the ICE BofA US High Yield Index’s interest rate exposure across rating categories and industry sectors.
- ICE US High Yield Best-in-Class ESG Index
This index eliminates both bonds of companies with significant involvement in controversial weapons and bonds of companies with high ESG Risk. It adjusts the weights of the remaining issues to match closely the ICE BofA US High Yield Index’s rating and industry sector distributions.
Index characteristics affect relative returns
Both proponents and critics of injecting ESG considerations into the investment process are keen to answer this question: Must ESG-conscious investors sacrifice return for the sake of their principles, or do they actually outperform standard indexes because companies achieve higher profits with less risk when they follow good environmental, social, and governance practices? To lay the groundwork to answer the question, the first portion of this analysis examines the ESG indexes’ composition for lurking variables that may affect that analysis.
Lurking - also known as confounding - variables figure prominently in our ongoing analysis of high-yield returns. Overlooking their impact can lead to incorrect conclusions that may prove costly to investors who formulate strategies based on the unrefined findings of historically based studies. In the present context, a high-yield investor may mistakenly embrace or reject an ESG strategy, based on historical performance vis-à-vis a standard high-yield index that in reality results from an incidental, non-ESG-related difference between the ESG index and the standard-index.
The table below shows how selective the three high-yield ESG indexes are by tabulating the percentages of the ICE BofA US High Yield Index’s issues that they retain. Selectivity increases from 72.7% for duration-matched ESG tilt to 80.7% retention for ESG tilt. Best-in-class, which excludes bonds of companies with poor ESG scores as well as those with heavy involvement in controversial weapons, is the most selective, retaining a minority (47.2%) of the standard index’s issues.
|
In the course of eliminating ESG-defined bad actors, the new indexes disproportionately exclude bonds rated lower than BB. Apparently, issuers of B and CCC or lower bonds are more involved than their higher-rated counterparts in activities and practices that run afoul of the ESG arbiters. The more selective an index is (moving down from tilt to best-in-class in the table), the higher is its issue-based concentration in the top rating tier.
All three ESG indexes have average ratings of BB3, one notch above the ICE BofA US High Yield Index’s B1. It is therefore appropriate to introduce into our analysis the ICE BofA BB US High Yield Index, which has an average rating yet one notch higher, i.e., BB2. Not surprisingly, in light of these facts, the table below shows that each ESG has an option-adjusted spread (OAS) in-between that of the overall high-yield index and the BB index’s spread. Like the rating agencies, the market rates the ESG portfolios more highly than the standard all-HY index.
Astute readers will exclaim at this point, “Wait! The descriptions of the three ESG indexes mention reweighting of the non-excluded bonds to match the ICE BofA US High Yield Index’s ratings distribution. The percentages in the table are based on numbers of issues. Are those issues not reweighted so as to keep the ESG indexes’ market-value-weighted BB concentrations at the same level as the standard high-yield index’s?”
The answer is that the reweighting process is evidently imperfect, perhaps as a result of pursuing multiple objectives. On June 30, 2020, the BB share of the ICE BofA US High Yield Index’s market value was 56.27%. All three ESG indexes’ BB shares were higher, at 65.16% for tilt, 65.45% for duration-matched tilt, and 57.95% for best-in-class. (Note that the ratings figures shown in the table are instead based on number of issues.)
|
Other incidental effects of eliminating approximately one-fifth to one-half of the ICE BofA US High Yield Index’s issues, depending on the ESG index, appear in measures of term risk. Compared with the all-high-yield index, the best-in-class ESG index has a meaningfully shorter average maturity, while the tilt portfolio has a modestly longer maturity. Note, however, that on the measure of effective duration, which is a more precise gauge of interest rate sensitivity than average maturity, deviations from the all-high-yield index are negligible and nonexistent, respectively, for the tilt and duration-matched tilt versions.
A final, important basis of comparing the standard high-yield index and the ESG versions is percentage concentration in Energy.
From Dec. 31, 2016 to June 30, 2020, the period in which we will compare ESG and standard high-yield performance, the ICE BofA US High Yield Energy Index had a mean monthly return of –0.128% with a standard deviation of 6.50%. That compares very unfavorably with 0.351% and 2.354% for the ICE BofA US High Yield Index. The substantial disparity in returns and volatility lends considerable importance to the tilt and duration-matched tilt indexes’ considerable underweighting of Energy, vis-à-vis the overall high-yield index, as detailed in the table below. Note, however, that as of June 30, 2020, the best-in-class index was perfectly matched with the standard high-yield index in terms of Energy’s share of total market value.
|
We have documented lurking variables that may significantly affect the performance comparison of ESG high-yield indexes with the standard high-yield index. Specifically, we have found potentially material divergences in ratings mix, term risk, spread, and concentration in the low-average-return (since 2016) and volatile Energy sector. These divergences will be important considerations in judging whether ESG-conscious investors have had to sacrifice return or have actually benefited by overweighting the bonds of ESG-defined good citizens. We now compare the standard high-yield index’s and the ESG indexes’ returns, both absolute and risk-adjusted.
Comparing conventional and ESG high-yield returns
Again, on June 17, 2020, ICE Data Indices, LLC launched three high-yield bond funds constructed with consideration of ESG factors. Return histories were provided from Dec. 31, 2016 forward. The newly available data enable us to address a central question in the debate surrounding ESG-based investing (see note 1): Must ESG-conscious investors sacrifice return for the sake of their principles, or do they in fact outperform standard indexes as a consequence of companies achieving higher profits with less risk when they follow good environmental, social, and governance practices?
The table below displays mean monthly returns, volatility (standard deviations), and risk-adjusted returns (Sharpe ratios) of the ICE BofA US High Yield Indexes, the ICE BofA BB US High Yield Index, and the above-listed ESG indexes. All three ESG indexes outperformed conventionally constructed high-yield indexes in both absolute and risk-adjusted terms during the available observation period. These findings are noteworthy, but do not definitively answer this piece’s central question.
|
To begin with, the ESG mean returns are not statistically different from the 0.351% generated by the all-high-yield index (hereinafter referred to as “All HY”). That is, the ESG indexes do not come close to achieving a 95% confidence level in a difference-of-means test. This finding implies that ESG-conscious investors do not have to sacrifice return for the sake of their principles, but the evidence does not demonstrate that adhering to ESG principles inherently enhances performance. (For anyone who was wondering, our finding of no statistical significance is not altered by analyzing quarterly, rather than monthly returns.)
Let us be fair regarding the superior ESG returns’ lack of statistical significance. The BB index’s total return premium over All HY is even greater than the ESG indexes’ yet it is not statistically significant, either. (Anomalously, from the standpoint of financial theory, BBs produce higher average returns than the all-high-yield index, which is riskier by virtue of also holding higher-default-probability Bs and CCCs.) To lapse momentarily into quantitative geekiness, the standard deviations are too great and the sample sizes are too small to establish significance for the data displayed in the table.
The BB results are important for a second reason. As our June 28 piece noted, the three ESG indexes have higher average Composite Ratings, at BB3, than All HY’s B1. It seems more than coincidental that the BB3 ESG indexes’ mean returns are in-between those of B1 All HY and the BB index, which has an average Composite Rating of BB2. Could differences in ratings mix represent a lurking variable? That is, observed superior returns on ESG indexes may in truth be attributable to some factor other than the good ESG scores awarded to the issuers represented in those indexes by Sustainalytics.
To explore that possibility, let us first note another finding displayed in the preceding table. The three ESG indexes achieved their total return edge over the ICE BofA US High Yield Index—statistically insignificant though it was—in the months in which the ICE BofA US High Yield Index posted negative price returns. That is, All HY’s 1.663% mean return in the up months surpassed all three ESG indexes’ mean returns, while its –0.842% mean return in the down months was exceeded by all three ESG indexes’ mean returns. (All of these differences, we note for the record, are not statistically significant.)
Sources of superior downside protection
Based on the foregoing evidence, one might argue that even if ESG-based bond selection does not deliver significantly superior average returns, it provides superior downside protection. That would be consistent with the notion that ESG bad actors pose exceptionally high event risk to investors, e.g., environmental disasters, boycotts initiated in reaction to antisocial behavior, or scandals arising from poor governance. In a selloff, one might reasonably conjecture, investors would be especially eager to exit obligations of these accidents waiting to happen. The question is whether the high-yield ESG funds’ lesser downside vis-à-vis the standard high-yield index is attributable to lower event risk or, rather, is a function of lower-risk ratings and industry profiles.
To address this question, let us now examine returns in March 2020. That is the down month with the biggest total return differential between the ICE BofA US High Yield Index and the ESG indexes since the latter's Dec. 31, 2019 inception. We focus on the comparison between All HY and the duration-matched ESG tilt index (hereinafter referred to as “DMESG”), which outperformed All HY by the greatest margin among the three ESG high-yield indexes in March 2020.
We noted earlier that as a side effect of excluding bonds of ESG bad actors, the three high-yield ESG indexes diverge from the All HY in ratings mix and term risk. For the duration-matched ESG tilt and best-in-class indexes, these effects are partially reduced by reweighting the retained issues with an eye toward matching certain characteristics of the parent, all-high-yield index, to the extent feasible. In addition, the ESG indexes have lower concentrations in Energy, which had a total return nearly twice as low as the next-worse-performing major high-yield industry in March 2020. Because DMESG’s interest rate sensitivity, as measured by effective duration, is no different from All HY’s, our search for lurking variables concentrates exclusively on ratings mix and Energy concentration.
As detailed in the table below, DMESG’s market value was more heavily concentrated in the lowest-risk BB segment (62.92% versus 51.11%) than All HY going into March 2020. Conversely, DMSEG’s concentration in the riskiest segment, CCC & Lower, was smaller than All HY’s (6.52% versus 11.59%). DMESG was also substantially less concentrated than All HY in the highly volatile Energy sector (4.58% versus 11.64%).
|
The lower portion of the table shows how the All HY portfolio would have performed in March 2020 if its ratings mix had matched DMSEG’s. At –10.984%, the reweighting of ratings would have improved its total return by 0.775 percentage points over All HY’s actual –11.759% figure. The ratings-matched version of All HY would not have equaled DMESG’s –9.295%, however. This means that DMSEG’s higher-quality ratings mix by itself does not explain its March 2020 outperformance of All HY.
Observe, however, that All HY’s return would have improved even more, to –10.000%, if instead of reweighting its ratings mix, we had matched its Energy concentration’s with DMESG’s. That improvement would have measured 1.759 percentage points. For sake of argument only, we note that adding that figure to the 0.775-percentage-point improvement from reweighting the rating categories produces a sum of 2.534 percentage points. That figure is nearly identical to the 2.464 percentage points by which DMSEG outperformed All HY in March 2020.
Adding the two variances together is not a completely accurate procedure, due to interaction among the factors. For example, the Energy weighting within DMSEG’s CCC & Lower component differs from the corresponding All HY. Adjusting weights for one factor does not hold the weight of every other relevant factor constant. Beyond that complication, numerous factors other than ratings mix and Energy concentration affect the comparative returns of DMSEG and All HY, notably differing concentrations in industries other than Energy. A full-blown performance attribution is a more extensive project than we propose to undertake in this piece. Even with the limited results derived from the table above, however, we can make a strong case that the ESG indexes’ superior downside protection derives in large measure from compositional features that are merely side effects of selecting issues on the basis of favorable ESG scores.
It does not automatically follow, however, that a high-yield investor who is indifferent to ESG considerations can replicate the superior downside protection of an ESG index simply by matching its ratings mix and Energy concentration. To test that proposition, we tabulated the percentage of DMESG bonds in each cell in the table below in March 2020. By way of illustration, DMG’s market-value-based concentration in CCC & Lower bonds, expressed as an integer, was 6%, split between 1% Energy and 5% Non-Energy.
|
We then created an equally weighted, ESG-blind portfolio by randomly selecting bonds from the ICE BofA US High Yield Index to fill each cell in the columns headed “Energy” and “Other.” If portfolio performance was solely a matter of sector allocation, we reasoned, then with a sample as large as 100 bonds, the randomly selected portfolio’s March 2020 return would closely approximate DMESG’s.
The March 2020 total return for our ESG-blind portfolio was –11.143%. That represented an improvement over All HY’s –11.759%, reinforcing the impression that DMESG’s superior downside protection in that month was at least partly attributable to its more conservative ratings mix and lower Energy concentration. The ESG-blind return, however, was considerably worse that DMESG’s –9.295%.
These results have important implications outside the present context of ESG-conscious indexes. They illustrate the difficulty of matching, much less beating, a benchmark over the short run, in a volatile period, merely by replicating the benchmark’s quality and sector profile. This is true even for a portfolio with more than three times as many securities as the 30 widely thought (albeit erroneously) to provide adequate diversification. In short, stock selection matters greatly when diversification is most needed, i.e., in a sharp downturn.
The results of our 100-bond-portfolio experiment leave open the possibility that favorable ESG scores contributed in part to DMESG’s superior downside protection during the severe March 2020 selloff. Further evidence on this point could be collected by segregating the All HY issues between those included and those excluded from DMSEG and comparing the two subsamples’ price returns, normalizing for differences in ratings mix and Energy concentration. That is a potentially fruitful topic for future research.
Research assistance by Lu Jiang and Zhiyuan Mei.
ICE BofA Index System data is used by permission. Copyright © 2020 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.
Notes
1. See, for example, Brad Cornell, “ESG concept has been overhyped and oversold,” Financial Times (July 17, 2020), p. 11. The Anderson Graduate School of Business, UCLA professor of climate change and finance contends that the story of higher returns for doing good is “backed by weak to non-existent evidence of promised pay-offs for either companies or investors, and fraught with internal inconsistencies that undermine its credibility." Along with Stern School of Business, NYU finance professor Aswath Damodaran, who contributed to the article, Cornell argues, “It is impossible to have an honest discussion about ESG when its advocates believe they occupy the moral high ground and view disagreement as immoral or unethical.”