SRI, data and bias: fund managers' bermuda triangle

However, the most striking challenges investors face is gaining access to data quality and relevance as well as managing the limits of these ESG data and the possible consequence on SRI strategy implementation.

50 SHADES OF DATA

ESG data include any indicators that shed light on the sustainability context of an asset, facility or company. Nowadays, four main types of ESG data can be defined: “Mandatory ESG disclosures”, such as financial filings, proxy statements or mandatory ESG extra-financial information, “Voluntary ESG disclosure” coming from sustainability reports, corporate websites, ESG disclosure surveys (i.e. CDP) or voluntary initiatives (TCFD, etc.). Furthermore, “Public and alternative sources” provided by various public sources such as the media, NGOs, governments and academics, and finally “Computed ESG information”. Computed data can derive on the one hand from ESG rating providers that use their proprietary methods to process and standardise existing data into a suite of metrics, scores, ratings and rankings. On the other hand, computed data come from more specialised providers that are covering specific issues (i.e. climate risk, social impact, etc.) Lately, we’ve also seen fintechs entering the game. As the number of company disclosures, filings and external sources have increased over time, an exponential mass of data continues to emerge, posing some undeniable questions for investors on how to fathom all this information, its materiality and how they can tackle ESG data limitations.

50 SHADES OF BIAS

These data could have a multitude of biases that could, if not correctly acknowledged and handled, put investors at risk and could lead to wrong investment decisions. Without being exhaustive, “bias in raw data” could be the first one. Indeed, ESG reporting is still voluntary, so neither metrics nor accounting methods provided are consistent, which can limit comparability across companies and sectors. This implies, among other things, a lot of missing data without clear reasoning (is a company not disclosing?), which can lead to a distortion in investors‘ analyses. “Sectorial bias” is another one, due to the fact that company-specific risks and differences in business models, are not always properly accounted for in composite ratings. Due to significant differences in business models and risk exposure, companies in the same sector are mostly assessed according to the same model.

Moreover, ESG data can carry a “geographical bias”, as regulatory reporting requirements and commercial standards on ESG disclosure vary considerably, causing important discrepancies between regions. European companies have, on average, better scores than US or Japanese ones, making global sectorial comparison and integration harder. Companies having stricter regulations on disclosure will be more in line with ESG rating inquiries. Furthermore, a “market cap bias” can occur, as higher market capitalisation tends to have significantly higher ESG ratings. Indeed, larger companies are providing more resources to answer third party questionnaires and develop a more nuanced and positive view of their activities. Therefore, a correlation between a company’s ability to produce ESG content and the quality of its ESG ratings could probably be established! “Cultural bias” also plays a major role.

FROM BACKWARD LOOKING TO FORWARD LOOKING

An SRI fund manager must be a conviction manager, and not a sheep manager who only looks in the rearview mirror.

Nowadays, ESG scores mainly have a backward-looking view, since rating providers’ reporting includes a time lag and is only updated on a yearly basis, which is not responsive in comparison to the financial timeframe. At best, ESG data can look at the present picture with the adding value of an ongoing investor engagement and controversy analysis. Nevertheless, beyond this timescale issue, today’s challenge in order to build responsible long-term strategies is to select the best ESG data that maximise financial performance and add a set of alternative data that makes it possible to anticipate the capture of a company’s weak signals to allow better reactivity.

Even though big data technologies will need to help deal with the exponential volume of ESG data, human beings remain key in ESG analysis, for instance in engagement activities with companies that enable support for the adoption of best practices but also by fostering new profiles and competencies in data science, capable of understanding these data, these methodologies, and translate them into innovative SRI strategies.