ESG is all the rage. Both large institutional and individual retail investors are increasingly demanding that the stewards of their savings demonstrate the consideration of Environmental, Social and Governance externalities in their decision-making. The aim is to ensure that investment returns are not generated at any cost, that risks associated with poor performance on ESG-related fronts are appropriately discounted, and that a higher cost of capital is imposed on the bad actors.
Similar to the credit rating agencies, third-party firms, such as MSCI, provide ESG ratings which can help fund managers factor these specific risks into their investment processes. For ESG focused investors, this all sounds great in principle, but how does it work in practice? Do these ESG scores really enhance the investment process? Aren’t these ESG risks already factored into an issuer’s credit ratings? Or, is the market already efficient at discounting ESG related risks such that there are no alpha opportunities from an ESG overlay on portfolio construction?
We have attempted to answer all of these questions through a quantitative approach that overlays MSCI ESG scores on the high yield asset class as represented by the USD component of the HW00 index (ICE BofAML Global High Yield).
ESG back testing methodology
As Figure 1 shows, MSCI coverage of the high yield asset class only began in 2006 and rated 50% of the US Dollar high yield market by the end of 2012, the point from which we have run the back tests. As we are attempting to ascertain the utility of ESG scores, we removed all unscored bonds from the index to create a custom ESG benchmark.
While there are numerous ways of systematically incorporating ESG criteria in a corporate bond back test, we have tested three approaches as described below:
ESG back testing results
The results of these back tests are presented in Table 1. To account for the active turnover, we apply a proprietary transaction cost model at each monthly rebalance point and present the post transaction cost results for each strategy to the right of Table 1. Furthermore, to facilitate a fair comparison to the ESG benchmark, we measure its turnover and apply the same transaction cost model in order to calculate its inherent post transaction cost return statistics.
We note that all the active ESG back-tested strategies have the same average rating as the ESG benchmark. They also have controlled average levels of duration which are close to the ESG benchmark’s duration. So, from a rating and duration risk perspective, these back-tested strategies are all comparable. A couple of points related to returns are worth noting from these results: (i) in two of the ESG constructions the portfolio yield is not overly compromised vis-à-vis the benchmark yield; and (ii) the gross returns of all three ESG back-tested strategies are higher than that of the benchmark.
When further scrutinising the post transaction cost results, we can make the following broad statements about the active ESG back-tested strategies relative to the custom benchmark: (i) the volatility of their returns is consistently lower; (ii) their risk-adjusted returns are consistently higher; and (iii) their maximum drawdown levels are significantly better.
Figure 2 shows the cumulative monthly returns for the ESG benchmark and for the three ESG back-tested strategies from 30th November 2012 to 31st July 2018. All returns are shown net of transactions costs, including the benchmark returns. The drawdown mitigation advantages of the ESG back-tested strategies are again quite evident in this chart. We can also easily visualise the outperformance of the negative filter strategy over the full period.
This study indicates that by either eliminating the lowest scoring credits from an ESG perspective (negative filter strategy), or by re-weighting the market portfolio based purely on ESG scores (ESG weighted strategy), an ESG overlay can be used to enhance portfolio returns. However, the ESG weighted back test also shows that in the real world transaction costs do matter. This favors the lower turnover negative screening approach. Given the near identical ratings profiles of all the back-tested strategies, and given that we specifically controlled for ratings in the ESG buffer strategy, the overall outcome suggests that MSCI scores are additive to traditional credit ratings. In other words, the contingent liabilities related to “E”, “S” and “G” are not necessarily factored into the rating agencies’ assigned ratings. So, at least as it pertains to the USD high yield corporate bond market, ESG scores can be utilized to enhance portfolio outcomes via lower drawdowns, reduced portfolio volatility and increased risk-adjusted returns. The results of this empirical work suggest that similar research should be conducted across other asset classes.
 MSCI provides our fixed income investment management team with an ESG data feed covering 6,000 global companies (13,000 issuers in total when including subsidiaries) mapping to over 590,000 equity and fixed income securities
 Agency ratings take into account a number of balance sheet and off balance sheet factors, and are an important input to the risk scaling processes that portfolio managers use when constructing portfolios. It is often assumed that ESG ratings are simply proxies for agency ratings – i.e. that a portfolio with a weighted average ESG rating of, say, BB has the same return and risk characteristics as a portfolio with the same weighted average agency rating. We look to see if this is actually the case.
 The ESG coverage statistics are in terms of market capitalization. ESG scores fundamentally describe the ESG attributes of a corporate entity, not an investment instrument. In the case of MSCI ESG data, most of the ESG scores fundamentally pertain to equity-issuing entities. So, as bond investors, our challenge is to correctly map from the bond-issuing entity to the correct equity parent. We have developed a proprietary mapping mechanism to ensure we are attaching the correct ESG scores to the benchmark bond issuers.
 MSCI ESG scores range from 0 to 10, where 0 represents the worst ESG quality and 10 represents the best ESG quality.
 We also maintain a 5% maximum issuer constraint in order to avoid issuer concentration risk.
 These net performance metrics are effectively proxies for those of a purely passive product constructed around this particular custom ESG benchmark.
 Most of this return outperformance is negated by transactions costs. In fact after costs have been applied, only the negative filter strategy’s return is above the ESG benchmark’s net return. This is partly due to the lower incremental turnover of this particular strategy when compared with the others.