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.
Note: This article should be read in conjunction with: "Fridson on Finance: ESG vs. broader high yield – assessing profiles & returns."
A prior article on environmental, social, and governance investing provides descriptions of three high-yield indexes engineered for sensitivity to high-yield ESG factors:
* ICE US High Yield ESG Tilt Index
* ICE US High Yield Duration-Matched ESG Tilt Index
* ICE US High Yield Best-in-Class ESG Index
All three of these indexes represent subsets of the issues that constitute the ICE BofA US High Yield Index. The index provider selects issues for inclusion based on ESG designations assigned by Sustainalytics.
New ESG conflict
The dispute over ESG-based investing, discussed in our previous article, has heated up in recent days. Leading asset managers, including Franklin Resources, State Street Corp. and Vanguard Group, sent letters asking the U.S. Department of Labor to reconsider investment rules floated by the department in June. The proposed regulations would require administrators of retirement plans to base their selection of investments solely on financial considerations. Administrators would be prohibited from subordinating return or increasing risk in pursuit of non-financial objectives (see note 1).
In their letter to the Labor Department, State Street Global Advisors, or SSGA, Deputy Global Chief Investment Officer Lori Heinel and General Counsel Katherine McKinley, described ESG factors as "drivers of long-term shareholder value for our clients" (see note 2). They contended that the Labor Department was conflating investment to advance a cause ("impact investing") with taking into account financial risks such as climate change and excessive executive compensation when selecting securities ("ESG integration"). SSGA's letter argued that pension plan participants would actually be financially disadvantaged by the proposed new rule as it would limit their access to a long-term, value-driven investment approach.
SSGA President and CEO Cyrus Taraporevala wrote in a Financial Times (London) opinion piece that his firm agrees with the Labor Department on an essential principle: Managers should not use pension plan assets to advance objectives that conflict with return maximization. He argued, however, that a growing body of research demonstrates the importance of combining ESG considerations with traditional financial analysis to maximize performance.
In particular, Taraporevala cited research that draws on State Street data and deals with companies characterized by good governance. High marks in the G of ESG reflected such traits as sound employee safety practices, effective supply chains and the agility to repurpose products to fulfill new market needs. The shares of such companies, Taraporevala stated, declined by less in March's equity market sell-off than those of companies with weaker ESG profiles (see note 3). SSGA's viewpoint contrasted with Brad Cornell's. The UCLA Anderson Graduate School of Management professor recently wrote in another FT opinion piece: "[T]here are costs to being good in many situations and denying these costs, or arguing that the benefits always exceed the costs, is dishonest" (see note 4).
The following analysis, focused on the high-yield bond market, addresses the central question that once again came to the fore in the recent intensification of the ESG debate: Must investors sacrifice return to follow ESG principles? If not, could it even be the case that restricting high-yield bond selection to issues with favorable ESG ratings actually improves performance?
Our prior study established the following key points that this week's article builds upon:
1. Casting out high-yield bonds with unfavorable ESG characteristics to create ESG-based indexes has the side-effect of reducing concentration in the lowest rating categories and in the volatile Energy sector.
2. From their Dec. 31, 2016, inception (launch date: June 17, 2020) through June 30, 2020, all three ESG-based indexes generated higher mean monthly returns than the "parent" ICE BofA US High Yield Index, with lower standard deviations and superior risk-adjusted returns (Sharpe ratios). The ESG-based indexes' returns, however, were not statistically different from the parent index's.
3. The ESG-based indexes total return edge was earned in down months, defined by negative price returns on the ICE BofA US High Yield Index. This raised the possibility that even if the ESG-based indexes did not outperform the parent index on a statistically significant basis over the long run, they genuinely provided investors superior downside protection during a sharp market downturn.
In a test of the proposition presented in point No. 3 above, our prior article examined relative performance in March 2020, the ICE BofA US High Yield Index's most negative total return month since the ESG-based indexes' inception. We found that the ICE US High Yield Duration-Matched ESG Tilt Index, or DMSEG, outperformed the parent, or ALL HY, index, -9.295% to -11.759%, in March 2020.
It was clear, however, that DMSEG benefited substantially from having lower concentrations than the ALL HY index in ≤ CCC issues, which returned 20.195%, and Energy, which returned -33.774%. Therefore, we could not conclude that DMSEG provided superior downside protection by virtue of companies with favorable ESG characteristics being inherently less risky than those with unfavorable ESG characteristics. We deferred to the present study the question of whether there was a residual ESG-related effect, in addition to low ≤ CCC and Energy concentrations, that helped to explain the ESG-based indexes' superior downside protection during March's severe downturn.
We worked at two levels in attempting to identify a residual ESG-related effect in the ESG-based indexes' March returns. The first involved the issue selection process of the ESG Tilt and Duration-Matched ESG Tilt indexes. Those two variants bend toward ESG objectives only by excluding bonds of issuers with significant exposure to controversial weapons. (Sustainalytics includes in this category anti-personnel mines, nuclear weapons, cluster weapons, biological and chemical weapons, depleted uranium and white phosphorus munitions.) The second level of our analysis focused on bonds with favorable ESG scores, as determined by Sustainalytics. These are included in the best-in-class index, while those with unfavorable ESG scores are excluded. In the discussion that follows, we refer to these two categories as Bad Citizens and Good Citizens.
Our objective was to determine whether involvement or noninvolvement in controversial weapons, or whether being a Good or a Bad Citizen in ESG terms, produced a statistically significant difference in the March mean returns of high-yield issues. Note that unlike the above-referenced returns of the indexes, which weight issues by market value, the following analysis weights issues equally.
We sought to eliminate, to the extent feasible, the possible impact of confounding factors. For starters, we removed all Energy issues from our samples, based on that sector's far-below-index return in March, as detailed above. Next, we addressed the issue of differences in ratings mix between the subsamples being compared, based on the abovementioned impact of ratings mix on the DMSEG-versus-ALL HY total return differential. Our method of dealing with that possible confounding factor was to break out the returns by Composite Ratings. These are the averages of Moody's, S&P Global Ratings and Fitch Ratings, as applicable, calculated by ICE Indices LLC.
In addition, we limited our samples to the largest capital structure priority category within the ALL HY index, the senior unsecured category. By doing so, we eliminated junior subordinated, subordinated and secured bonds as well as certain bank issues with specialized seniority designations. This procedure prevented noncomparability that might arise from the fact that default probability and loss given default differ between like-rated bonds of different seniority levels (see note 5).
We did not attempt to limit differences in duration between subsamples, one final factor that could potentially be confounding. Doing so would have required further removals of issues — those with very long and very short maturities — on top of the previous culling that was already producing problematically small sample sizes at some rating levels. Note, however, that the Duration-Matched ESG Tilt Index, like the other two ESG-based indexes, outperformed the ALL HY index. This suggests that adjusting for differences in interest rate risk would not alter our conclusions.
In each comparison, whether of full samples or rating-defined subsamples, we conducted a difference-of-means test. Our threshold for affirming statistical significance was the combination of a t-statistic with an absolute value of 1.96 or greater and a p-value of 0.05 or less, indicating a 95% confidence level. Note, however, that in the first table below, we counted the borderline results at one rating level, BB3, as a statistically significant case.
Involvement in controversial weapons
The table immediately below shows a higher mean return for bonds of issuers with significant involvement in controversial weapons than for bonds of issuers with no such involvement. On its face, this finding contradicts any assertion that companies in the first category are inherently riskier than those in the second category. In fact, it implies the opposite. The 0.84-percentage-point difference in mean returns, however, does not come close to meeting our test of statistical significance.
Curiously, six rating-level comparisons do show statistically significant differences in means. Those cases are identified by shading in the Difference column. Even more surprisingly, those six cases are divided between two in which issues with controversial weapons involvement outperformed those without such involvement and four in which the opposite occurred.
We believe these strange-looking results represent statistical noise. In five out of six cases, one or both subsamples has an issue count of less than 30. That is a common, if not rigorously supported, threshold for defining an adequate scientific sample. In the fifth case, one subsample's issue count is barely above the rule-of-thumb threshold, at 31.
In summary, we find no evidence that high-yield investors can obtain superior downside protection by concentrating in bonds of companies that lack significant involvement in anti-personnel mines, nuclear weapons, cluster weapons, biological and chemical weapons, depleted uranium and white phosphorus munitions. On the other hand, the lack of statistical significance in the bottom line of the preceding table indicates that investors sacrifice no downside protection by avoiding such issues. For some investors, knowing that they will incur no financial penalty will be sufficient to induce them to refrain from providing financial support to the production of these controversial weapons. Still, other investors will be willing to forgo some return for the sake of their principles. Investment regulations, however, may bar fiduciaries from adopting such a course.
Good and bad ESG scores
The table below shows that ESG Good Citizens, as defined above, outperformed ESG Bad Citizens in March by a statistically significant margin of 1.68 percentage points. On the face of it, high-yield investors were actually rewarded for high-mindedly restricting their holdings to bonds of companies with commendable environmental, social and governance practices. Examination of the underlying data casts doubt on that conclusion, however.
To begin with, the differences in means in the majority of the rating-based comparisons are not statistically significant. At the B2 level, the direction of the difference is not only statistically insignificant but in the wrong direction, i.e., the Bad Citizens outperformed the Good Citizens. These results cannot be explained away by inadequate sample sizes, given the respective issue counts of 41 and 64. Also with adequate sample sizes, B3 Good and Bad Citizens generated nearly equivalent mean returns.
One additional analysis clinches the case, in our judgment, that the Good versus Bad Citizens difference in means is a case of spurious statistical significance. We note that 12.7% of the Bad Citizens but only 4.1% of the Good Citizens are rated CCC1 or lower. Although we attempted to eliminate credit quality differences as a confounding factor by stratifying returns by Composite Rating, that procedure did not entirely solve the problem. Inspection of the Difference column leads to a supposition that the figures displayed for the CCC1, CCC2 and CC categories are outliers that undermine the statistical integrity of the analysis.
Testing just the issues rated B3 or higher produces means of -9.39% and -8.81%, respectively, for the Bad and Good Citizens. The respective standard deviations are 7.43% and 7.97%. At -0.58 percentage point, the difference in means is not statistically significant (t-statistic = 0.96, p-value = -0.34). Mirroring our finding in the Controversial Weapons comparison, we conclude that high-yield investors are neither rewarded nor penalized for being on the side of the angels on a full array of environmental, social, and governance matters.
Based on the relatively short, 3.5-year history of the relevant indexes, high-yield investors need not sacrifice return to adhere to investment rules that exclude bonds of issuers they find objectionable on environmental, social and governance grounds. This statement assumes, however, that returns are not reduced by lowering the number of purchase candidates. Such constraints can pose difficulties during the high-yield market's not infrequent bouts of supply scarcity, when portfolio managers find it difficult to put cash to work at palatable yields. Periodic supply shortages are especially problematic for mutual fund managers, who struggle during such episodes to invest large inflows.
Keeping large amounts in cash equivalents while awaiting better opportunities is not a viable option for those managers due to investor pressure to maximize current yield.
Our findings do not support, in the case of high-yield bonds, the contention of some proponents that concentrating on issues with favorable ESG scores increases investment performance. If issuers devoid of significant involvement in controversial weapons, or that have favorable ESG scores, are inherently less risky than like-rated issuers that lack those characteristics, their bonds ought to beat the conventional high-yield index most handily in periods of sharply rising risk aversion. Our tests in one such period (March 2020), however, found no statistically significant performance differential attributable to ESG factors. The ESG-based high-yield indexes' outperformance of the ALL HY index in that month was rather a function of relatively low exposure to ≤ CCC and Energy issues, a side-effect of excluding issues on ESG-related grounds.
Research assistance by Lu Jiang and Zhiyuan Mei. Additional research assistance by Daniel Navaei.
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.
1. Ross Kerber, "Top fund firms oppose U.S. roadblock to green retirement funds," Reuters (July 31, 2020).
2. Attracta Mooney, "State Street lashes out at sustainable investing rule," Financial Times (Aug. 3, 2020), pp. 1-2.
3. Cyrus Taraporevala, "U.S. regulators wrong to dismiss ESG investing for pensions," Financial Times (Aug. 3, 2020), p. 17.
4. Brad Cornell, "ESG concept has been overhyped and oversold," Financial Times (July 17, 2020), p. 11.
5. See "Valuing like-rated secured, senior unsecured bonds," (LCD News, Aug. 31, 2016). Synopsis: Counterintuitively for many market participants, secured bonds have wider spreads than like-rated senior unsecured bonds. That relationship correctly reflects comparative risk.