ESG integration comes with a significant and evolving challenge: ESG data. No matter the investment style, proper implementation requires reliable and current data to both construct ESG portfolios and confirm they’ve succeeded in meeting investors’ ESG objectives. For quantitative investment methodologies, ESG data must also be comprehensive and have an extensive history.
In this blog, we’ll cover some of the requirements for useful ESG data, and how current ESG data sources are lacking.
At the simplest level, data reliability means that the data sources must be trustworthy, i.e., they have low frequency of errors and high internal consistency (e.g., changes in pillar scores are reflected in overall composite ESG scores). Ideally, the metrics should also be clearly defined such that the figures can be independently verified, at least in principle. However, at both the simplest and more demanding levels, ESG data present severe challenges.
For one thing, as investors’ interest in ESG is booming, new databases are regularly introduced, while older databases continue to evolve. This often negatively impacts the consistency of the data.
Also, unlike standard company fundamentals which have been reported and analyzed for many decades, there’s still little consensus on how to define or quantify ESG data. Even within the dataset of any single vendor, there are often inconsistencies due to changes in the raw-data sources, calibration methodologies, or ad hoc manual adjustments.
Finally, ESG data are unavoidably subjective to a large degree: all three pillars (Environmental, Social, and Governance) reflect implicit value systems, and their quantification depends on each provider’s judgement regarding which issues to consider, how to calibrate the rating scale, or how to distill the underlying ratings to a small number of figures. This opens the door to further inconsistencies at each step of this manual process.
For non-ESG types of data, the time interval between news or company announcements and when they are accurately reflected by data vendors has been generally compressed to a matter of days or hours. This is not the case for ESG data, partly because of the amount of manual labor involved in procuring the data, cleaning them up, or curating them in other ways; also because the impact of news can be hard to ascertain at the time it becomes public knowledge. This results in analysts updating ESG data on a weekly or monthly basis, at best. And even at this low frequency, it is often the case that analysts revisit only a fraction of the companies in the database at each update, so that many months (sometimes more than a year) are required for a full refresh of the database.
ESG data are difficult to access at high quality for broad index universes. Ratings for typical multidimensional ratings models are very resource intensive for this many securities because of the lack of similarity in the raw data sources (e.g., company filings, news, regulatory action, etc.) and the thorough cleaning required. Usually, only larger data vendors can afford to support the large teams that need to be employed for such a task.
Normalization poses special challenges for ESG data, and it can have unintended consequences. For example, if normalization does not include an industry adjustment, specific industries will often have a uniform and persistent poor rating, and there will be little incentive for companies within these industries to improve. Also, if normalization does not result in a compact range of scores, then a few companies will exhibit such extremely high or extremely low scores that including or excluding them respectively can be used to ‘greenwash’ portfolios through quite small tweaks in the holdings that don’t materially affect the ESG profile, but which nevertheless result in big shifts in the headline figures.
ESG considerations have only been the focus of broad attention for a short period of time, so there is a lack of historical datasets that can be reliably used to extend the present rating systems far into the past. Subjective aspects of the data, which require a high degree of manual intervention, exacerbate this challenge which is both costly to apply over large datasets and hard to do in a way that ensures internal consistency.
Furthermore, ESG considerations by their very nature tend to evolve considerably over time. For example, the rating of a company’s environmental credentials is unavoidably affected by the prevailing consensus at the time as to what issues materially affect the environment, by technological progress, and by regulations that can dramatically shift the boundaries of what is legally permissible, etc.
ESG data may have its share of issues, but we believe there are some ways to mitigate them. Learn more in our latest paper, Overcoming ESG Data Challenges.
The views presented are for general informational purposes only and are not intended as investment advice, as an offer or solicitation of an offer to sell or buy, or as an endorsement, recommendation, or sponsorship of any company, security, advisory service, or fund. Nor do they purport to address the financial objectives or specific investment needs of any individual reader, investor, or organization. The views are subject to change at any time based upon market or other conditions, are current as of the date indicated, and may be superseded by subsequent market events or other conditions. The information, analyses, and/or opinions expressed are not intended to provide any specific financial, economic, tax, legal, investment advice, or recommendations for any investor. It should not be relied on as the sole basis for investment decisions. While every attempt is made to ensure that all information is accurate, there is no representation or warranty, express or implied, as to the accuracy and completeness of the statements or any information contained herein. Any liability therefore (including in respect of direct, indirect, or consequential loss or damage) is expressly disclaimed. Past performance is no guarantee of future results. Investing involves risk, including fluctuation in value, the possible loss of principal, and total loss of investment.