Assessing the impact of ESG in fixed income
New academic research suggests that enhancing the ESG profile of a European fixed income portfolio would not have undermined performance over the last 10 years, with tilts improving performance at the margin and controversial sectors making little impact historically. Insight believes this provides additional support for considering ESG risks in fixed income portfolios but reinforces the importance of understanding the purpose and limitations of using ESG data.
Key results and investment implications
At a minimum, the study found that enhancing the ESG profile of a fixed income portfolio has not led to any deterioration in the risk-adjusted performance of fixed income portfolios. Given this, investors should feel confident that integrating ESG considerations could be a worthwhile exercise.
Key investor takeaways
The study provides additional independent, academic support for the consideration of ESG risks in fixed income portfolios, gives investors a better understanding of the influences on performance and highlights that there are still many areas where further research would be helpful.
Insight believes the study also reinforces the importance of understanding the purpose and limitations of using ESG data – such as coverage gaps, varying quality and different methodologies – and highlights the need for ESG data to mature at a much faster pace than it has done to date.
Fixed income assets are the core allocation for many institutional investors. However, much of the academic research into how environmental, social and governance (ESG) factors influence investment performance has focused on listed equity markets rather than fixed income.
While the industry has a feeling for how ESG factors impact performance in fixed income, Insight felt it was important to commission a rigorous, independent academic study on the topic.
Rather than seeking to identify how much out or underperformance an ESG strategy would deliver – which would lead to spurious conclusions as the results would be dependent on the time period, market, ESG data source and method of implementation – Insight asked Bayes Business School (Bayes) an open question to ‘assess the implications of applying ESG factors within a fixed income portfolio’.
In particular, they:
- Conducted detailed analysis on the benefits or otherwise of integrating ESG considerations into a European corporate bond portfolio
- Explored how different ESG implementation strategies affected performance
As a basis, Bayes created a hypothetical ESG reference portfolio (‘benchmark’ for the purpose of this study) based on the constituents of the iBoxx EUR Corporates bond index sourced from IHS Markit1 , overlaid with corporate ESG data from Refinitiv 2.
Using 10-year data to 31 December 2021, Bayes analysed the impact on performance of three portfolio construction methods:
- Quintile scores: they ranked the universe into quintiles using composite ESG scores; individual E, S and G scores; and ten sub-categories, such as emissions. They also separated sector effects, caused by the inadvertent deviation to sector weights relative to the benchmark, to isolate the ESG effects
- Portfolio tilts: they adjusted bond weights up or down relative to the benchmark based on their ESG scores, using two industry-standard tilting methodologies
- Controversial sector exclusions: they excluded the most common controversial sectors – tobacco, mining, defence, and oil and gas producers – in isolation and in aggregate
They also examined the impact of ESG scores on tail risk.
Higher ESG ratings improved risk-adjusted returns, but beware the many subtleties
Bonds with the best aggregate ESG scores outperformed bonds with the worst ESG scores
First, Bayes split the universe of bonds into quintiles, based on each issue’s composite ESG score. They found that the top 20% of bonds by ESG score outperformed the bottom 20% by roughly 3% – a statistically significant margin – over the decade studied (see Figure 1).
An ESG effect was visible at the aggregate level
Could the outperformance of the top-quintile portfolio be due to inadvertent tilting towards higher-performing sectors? To investigate, Bayes separated out the ‘sector effect’ (by adjusting the sector exposure of each quintile to match the sector exposure of the benchmark) to reveal the ‘ESG effect’.
- Top quintile: As illustrated in Figure 2, some of the historical outperformance relative to the benchmark is attributable to the sector effect, but a larger amount comes from the ESG effect.
- Bottom quintile: All the benchmark-relative outperformance is due to the sector effect with the ESG effect being negative.
There are many subtleties which are missed when just comparing top and bottom quintiles based on aggregate scores
Such a broad analysis does not always tell the whole story, as illustrated by the following observations which were uncovered after further interrogation:
- When looking at aggregate ESG scores, there is not a smooth increase in average return from quintile 5 (the worst ratings) to quintile 1, with quintile 5 outperforming quintiles 2-4. Also, while the top quintile outperformed the bottom quintile, both segments outperformed the benchmark index.
- Outperformance was not consistent over the whole period, being largely concentrated to the three years to early 2016, and between late 2018 to early 2020.
- The results were mixed when focusing on the individual E, S and G factors. As an example, ranking by environmental scores alone, top-quintile portfolios outperformed the bottom quintile, but the cumulative sector-adjusted performance for the bottom quintile was higher than for the top quintile. The results using just social scores were mixed and the results focusing on governance alone was muted.
- Even at the sub-category level, there were some very disappointing results. For example, the study found that those companies with the highest scores for human rights significantly underperformed the benchmark with almost 100% statistical certainty.
- The results suggest that pursuing higher ESG-scoring portfolios could lead to more attractive risk-adjusted returns
- However, over short time periods, the exposure to issuers with higher ESG ratings may not lead to a return benefit, and may potentially detract performance
- Focusing exclusively on a single ESG measure or sub-category is not a reliable indicator of superior performance, potentially due to the introduction of sector concentration risks.
ESG tilts would have historically improved performance at the margin
Excluding low-ranking issues had little impact or is impractical
One way to increase the ESG scores of portfolios is to exclude low-ranking issues. The study found that excluding the bottom quintile of ESG scoring issuers did not significantly affect past performance. It also found that unless investors only focused on the top 20% of bonds by ESG score, there was no discernible difference in outcome – outperformance in this case.
Excluding 80% of securities in a fixed income benchmark is not a practical approach due to liquidity and default risks that would become far more acute with such a concentrated portfolio.
Tilting a portfolio to favour stronger ESG scores positively influenced the return at the margin
A more popular way to increase the ESG scores of a portfolio is to tilt individual bond weights up or down relative to the benchmark based on their ESG scores. Using a tilting formula similar to that used by MSCI in some of its Factor Index range but applied to aggregate ESG scores, Figure 3 shows some historical marginal outperformance.
Similar results were observed when tilting scores based on individual social and governance scores, and by sub-category labels. The results were more positive when evaluating performance based solely on the environmental scores.
The results were also similar when using the tilting methodology adopted by FTSE, suggesting that the tilting methodology employed did not influence the result.
- Portfolios can be tilted towards particular ESG characteristics without having a material effect on the risk and return characteristics of the portfolios
- Results suggest that the choice of tilting methodology did not affect the outcome
Excluding controversial sectors would not have hurt historical performance
Exclusions are a common approach to addressing ESG issues. Insight note that such an approach is typically taken to reduce the wider costs for society or the environment (‘externalities’) of entities held within a portfolio, rather than to reduce financial risks associated with ESG issues.
The historical returns from a portfolio that excluded the most common controversial sectors – tobacco, mining, defence, and oil and gas producers – were statistically identical to the benchmark (see Figure 4). The main reason is that these sectors only account for a small proportion of the benchmark.
This is a positive finding for investors that use an exclusionary approach, as it shows that, in this sample over the past 10 years, such an approach would not have detracted from investment returns.
- Investors can align their portfolios with their ESG views through the exclusion of controversial sectors without impacting performance unless a large proportion of the benchmark is excluded
Enhancing ESG credentials generally led to a reduction in the tail risk of a portfolio
One of the supposed advantages of integrating ESG considerations into a portfolio is that it has a positive impact on tail risk. To investigate this, Bayes examined a range of risk metrics that focus on the measurement of tail risk.
Figure 5 shows the figures for quintile portfolios based on composite ESG scores. Generally, there is an increase in tail risk as we move from quintiles 1 towards quintile 5. Bayes observed similar results when looking at environmental and social scores, but not when focusing on governance scores.
While it is difficult to draw definite conclusions from the results with regard to the relationship between ESG ratings and downside risks in a corporate bond portfolio, arguably the results show that ranking using the composite ESG or individual environmental rating generally leads to a reduction in downside risk.
- Enhancing the ESG credentials can help to reduce the frequency of extreme downside outcomes
Key investor takeaways
Insight believes the study reinforces the importance of understanding the purpose and limitations of using ESG data, and the need for ESG data to mature at a much faster pace than it has done to date.
Determining the purpose of ESG ratings
It’s important to determine whether any chosen ESG ratings or scores aim to reflect financial risks, such as how an entity is exposed to and manages ESG issues that could affect their creditworthiness; or broader non-financial performance of an entity regarding ESG matters, such as the environmental impact of its operations.
In Insight’s experience, ESG ratings in most cases are more focused on financial risks, but there is a variety of approaches taken by different providers today.
Understanding the limitations of ESG data
Major limitations of ESG ratings include:
- Coverage gaps: Not all data providers will have the same coverage on each and every issue, creating data gaps. Also, the choice of any ESG benchmark used will influence the results
- Varying quality: Even when data providers have ESG data on issues, there may be weaknesses due to incomplete, irrelevant or stale data
- Different methodologies: Different data providers can have widely differing ESG ratings for the same issuer, while differences are typically much narrower for credit ratings
Insight illustrate the last point in Figure 6, taken from an OECD study published in March 2022. For example, one provider gave Boeing a rating around 30 while another a rating around 90 (on a standardised scale).
ESG ratings agencies may produce very different answers because they employ very different methodologies. By contrast, credit rating agencies have the more precisely defined task of calculating the probability of default, meaning methodologies are more aligned and ratings are typically bunched far closer together.
Addressing the data limitations
Below are some ways that Insight is trying to address the data limitations:
- Clarity on usage: Insight’s proprietary Prime ESG risk ratings aim to highlight where issuers are exposed to ESG risks that could affect their creditworthiness, rather than to score issuers on their non-financial performance or activity regarding ESG issues.
- Multiple ESG data sources: Insight’s proprietary Prime ESG ratings source data from multiple ESG data providers to broaden the coverage.
- Qualitative overlay: Insight overlay third-party data with qualitative input from their in-house experts to strive for a more accurate and reliable reflection of the risks that issuers face. Insight believe their judgement and experience, coupled with direct and collaborative engagement, helps to transform the data from information into insights useful for decision making.
- One input into a broader process: The ratings from Prime form an additional input into Insight’s investment process and are not the only driver of decisions.
- Addressing data gaps: Insight use proprietary questionnaires to help fill data gaps and, where opportunities arise, seek to engage more widely to encourage more transparency. This is particularly important in asset classes such as secured finance, where the quantity and quality of ESG disclosure lags more traditional credit markets.
While the above initiatives help, Insight are striving to do more in these areas and beyond. Insight also believe that the industry needs greater regulation – to drive the provision of standardised data that is most relevant to decision making, and is reliable and accurate – and greater independent verification of the validity of what is published to hold ratings providers to account.
While Insight are encouraged by recent moves, and planned initiatives, to bring such developments about, ESG data needs to mature at a much faster pace than it has done to date.
This report was written by Insight Investment. As such, it is in its voice as opposed to that of BNY Mellon Investment Management.
The value of investments can fall. Investors may not get back the amount invested. Income from investments may vary and is not guaranteed.
1 The iBoxx EUR Corporates bond index is the property of Markit Indices GmbH and/or its affiliates (“Markit”). iBoxx™ is trademark of Markit. Markit is an independent third party provider of broad and investable indices. Any investment funds or vehicle or investment strategies referred to in the Research are not sponsored, endorsed, sold, or promoted by Markit. Markit does not make any warranties or representations on the accuracy, fitness or purpose or results to be obtained by using the above index and disclaims all liabilities in this regard. The views and opinions of the authors of the research are his/her own and may not necessarily represent the views or opinions of Markit or any of its affiliates. The index provided by IHS Markit is subject to disclaimer currently available here (and as updated by Markit from time to time): https://ihsmarkit.com/Legal/disclaimers.html.
2 Model results have certain inherent limitations. Unlike an actual performance record, model results do not represent actual trading/returns and may not reflect the impact that material economic/market factors might have. Clients’ actual results may be materially different than the model results presented. Please see the full study for limitations identified by the study.
3 Source: Calculations by Bayes Business School, based on data from Refinitiv and IHS Markit
4 Source: Calculations by Bayes Business School, based on data from Refinitiv and IHS Markit
5 Source: Calculations by Bayes Business School, based on data from Refinitiv and IHS Markit
6 Source: Calculations by Bayes Business School, based on data from Refinitiv and IHS Markit
7 Source: Calculations by Bayes Business School, based on data from Refinitiv and IHS Markit. VaR 95 (Value at Risk at the 95% confidence level) is the return below which the bottom 5% of returns fall, and Con VaR 95 (Conditional VaR) is the average of those returns. Downside Deviation is the standard deviation of the negative monthly returns. The Sortino (Ratio) is the average excess return of the portfolio divided by the Downside Deviation.
8 Source: Refinitiv, Bloomberg, MSCI, Yahoo Finance, Moody’s, Fitch, S&P and OECD calculations. As at 31 March 2022.
1055750 Exp: 01 January 2023