With contribution from Robert E. Furdak and Jason Mitchell

Introduction

Incorporating ESG analysis has been the natural habitat for sovereign bond investors for decades – especially as it pertains to governance. What is new is that ESG data have matured over the last decade, and we are entering a phase where the data have both a long-enough history and broad-enough coverage to make it interesting to quantitative investment firms.

At Man Group, we have been able to demonstrate alpha from ESG factors in systematic security selection for equities. Can the same be said for sovereign bond selection? After all, ESG analysis – particularly the ‘E’ & ‘S’ components – have become increasingly important in forming long-term growth, inflation and monetary policy outlooks, all of which are important factors investors look at for sovereign bond selection.

For the reader in a hurry: we have not been able to find systematic alpha in ESG data for sovereigns so far. However, for a discretionary fund manager, we believe ESG data analysis is a potential rich area for idiosyncratic alpha.

The link between ESG data and discretionary analysis for sovereigns

The alignment of investment returns and long-term sustainability objectives has been the subject of many finance researchers for several years, not least at Man Group.[1] 

Naturally, good – or improving – ESG factors can help the investment process, and gauge the solidity of the tax base for a sovereign issuer, and the sustainability of government finances as it relates to E, S and G. The reverse also applies: taken to the extreme, it is unprofitable to invest in a security issued by an organisation that goes bankrupt due to bad management and/or running the business or country in an unsustainable way.

We believe it is worth looking at the correlations of the E, S and G scores across vendors to determine if there are any areas of data confusion that can provide idiosyncratic alpha.

As Figure 1 illustrates, there exists a very high level of agreement across two well-known ESG data vendors in both governance and social data for sovereign bonds. This suggests to us that the data is well known, not subject to much debate, and thus potentially lacking alpha. What is interesting, though, is that there is very little agreement at the ‘E’ level.

Figure 1: MSCI, Sustainalytics Agree on ‘S’ and ‘G’ Data for Sovereign Bonds; Less So for ‘E’

Source: MSCI, Sustainalytics; as of 21 March 2022

The pronounced disagreement on ‘E’ scores is worth analysing. At a headline level, differences are already obvious, with MSCI defining ‘E’ simply as environment, whereas Sustainalytics uses a broader ‘natural and produced capital’ category. Under this heading, Sustainalytics considers ‘corruption’ and ‘rule of law’ as ‘E’ – rather than the more traditional classification under ‘G’.

More detailed analysis reveals more subtle differences in the construction of the ‘E’ scores, whereas the frameworks surrounding ‘S’ and ‘G’ are strikingly similar. For example, MSCI focuses on environmental externalities, whereas Sustainalytics more directly considers factors like carbon intensity. This direct versus indirect approach appears throughout the analysis framework, with another example of Sustainalytics using ‘land below 5 metres’ to represent sensitivity to a changing climate, whereas MSCI less prominently adds ‘vulnerability to environmental events’ as a general category.

So, using aggregate scores could point to limited alpha potential if only using aggregate ESG scores. However, there is more scope for alpha when considering ‘E’, ‘S’ and ‘G’ scores and the various components separately.

Overall, though, there is a positive correlation between credit ratings and ESG scores[2] (Figure 2), which implies that: (1) ESG data has analytical value for sovereign default probability; and that (2) this value is already priced in, to some degree.

Figure 2: Credit Ratings Are Positively Correlated to ESG Scores

Source: Bloomberg, Sustainalytics; as of 21 March 2022

Note: Credit ratings range from 1 for AAA to 19 for CCC3; the lower the Sustainalytics ESG score, the better.

That is not the same as saying that ESG data are not additive – it is quite possible that ESG data are additive after a quantitative transformation that removes the fundamental data content that is captured in price data. This could, for instance, be done through regressions.

Researchers at Barclays[3] have done such an exercise and found no evidence of alpha in systematic selection of sovereign bonds using ESG data, even after regressing out fundamental factors such as credit ratings. However, they did conclude that the ESG score of a portfolio of sovereign bonds could be optimised without sacrificing return beta, or generating negative alpha, which may imply the ability to improve risk-adjusted returns by lowering the likelihood of experiencing future negative shocks through ESG data.

Overall, we believe ESG analysis for a discretionary fund manager can be additive to sovereign debt selection in at least three ways:

1. A discretionary fund manager can disaggregate, dissect and analyse the data in order to reach a conclusion that disagrees with the consensus;
2. While aggregate ESG scores agree across data providers, there is more debate and disagreement at more granular levels of data – especially for the E pillar – hence potentially more scope for alpha;
3. Importantly, increased investor engagement, net fund inflows and regulatory changes may benefit investments with better ESG scores in the future.

Systematic alpha in ESG data for sovereigns

In our article ‘ESG Data: Building a Solid Foundation’, we demonstrated that there was alpha to be found from ESG factors in systematic security selection for equities. Is there an alpha signal in ESG data for systematic selection of sovereign debt issuers too?

Before embarking on a quantitative quest, we have some thoughts on this from the outset.

First, positive ESG scores can be an indication of a corporate or country’s ability to grow in a stable and sustainable way. While an equity investor can benefit from profitable growth, this may be less relevant for debt securities. A large part of successful debt investing is more about accurately identifying the probability of default than the market, i.e., selecting between sovereigns that pay their coupons and return the principal at maturity versus those that are likely to become distressed.

Second, we would posit that historically, there has not been alpha in ESG data as it pertains to developed market sovereigns with strong ESG characteristics, especially if the data confirm common knowledge. Recently, however, this pattern may have changed as environmental data is increasingly being used to drive fiscal[4] and monetary policies.[5] There could, of course, be alpha in weak ESG data for sovereigns, to the extent that ESG factors could lead to negative shocks such as natural disaster incidents that a country is ill-prepared for, necessitating large fiscal outlays that could undermine the ability to pay coupons or principal. 

Third, a backtest of historical ESG data may well be a poor guide to the future, as environmental and social considerations were not perceived to be as acute or important as they are now in the eyes of citizens, governments and investors.

In our analysis, we were not able to find systematic alpha in ESG data for sovereigns – at least not yet. This could be due to the following reasons:

• First, contrary to ESG data for equities, where there is much disagreement between data providers, ESG data for sovereigns agree with each other at the aggregate level – as well as for S and G factors. This suggests there is little controversy and debate, especially for developed-market (‘DM’) sovereign issuers. Low controversy implies low alpha content;
• Second, we find there is a high correlation between ESG data and fundamental factors such as credit ratings. This is possibly a function of the large influence of the ‘G’ component in ESG scores, which has long been part of sovereign bond analysis;
• Third, environmental and social constraints on global production have been less significant in the past, when they were non-binding and less dominant, while macro factors dominated. Thus, a test of historical ESG data is not likely to show as much relevance as one might expect going forward. Increases in pollution, population and climate events have resulted in more economic sensitivity to environmental and social factors.

Conclusion

We have not yet found strong evidence that there is alpha in using headline ESG data for sovereign security selection in a systematic way. 

However, an ESG approach to sovereign debt selection can be additive in at least three ways: first, a discretionary fund manager can disagree with the data and the consensus, due to the edge the fund manager and their process provide; second, while aggregate ESG scores are in agreement across data providers, there is more debate and disagreement at more granular levels of data, hence potentially more scope for alpha; and third, flows and regulatory developments are likely to benefit investments with better ESG scores in the future.