As explained in previous posts in this series (again, disclaimer made in the first post still stands!), investors may be better versed with the key issues that Contemporaries focus on, and are given more up-to-date, standardized and reliable ESG ratings for companies. However, data scores given are more volatile when comparing Contemporaries with Traditionalists. When comparing TVL with MSCI, TVL has higher standard deviation and variance in ratings, with standard deviations of 1.41 and 0.97 respectively.
However, this is not to say that this volatility is bad. Humans have a habit of lowering entropy or seeking out patterns, some of which may not entirely exist. In the AI world, it is purely algorithmic, determined by a machine. There is no conscious decision making that is forcing the AI algorithm to lower deviation in data. It simply uses the information at hand, does some calculations, and comes up with a rating.
This is a good thing that reinforces the objectivity of Contemporary approaches. Nevertheless, it is not to say that the Contemporary approach is entirely perfect. AI algorithms are highly dependent on large volumes of data to enhance learning and provide outputs closer to the “true value” we are looking for. Therefore, Contemporaries would definitely need lots more data to help fine tune their system. For example, data is not evenly distributed geographically: there is more data to be scrapped in the developed world than there is in developing countries. There is no equality in data access, and things such as data censorship, government suppression of information, or simply a lack of infrastructure to relay data online can all affect the reliability of the AI algorithm used by Contemporaries.
Moreover, Contemporaries are completely dependent on external sources of data. It is important that there are data vendors available to them, or rather more data vendors may be needed. But more global data sources are becoming increasingly available, and may be used in the near future to help enhance the Contemporary method of ESG rating. Two examples include the Global Forest Watch (GFW) and MethaneSAT. GFW provides forest data globally, using satellite data as well as a network of partners to collect this information. This data is publicly available and free to use. MethaneSAT is a Jeff Bezos-funded initiative utilizing satellites to collect methane data all around the world and at a faster rate than scientists can measure from Earth. Data is publicly available to see how companies and governments are progressing with their carbon initiatives.
MethaneSAT boasts several impressive feats. Its satellites can identify emissions across the globe and measure the amount of emissions released from pre-determined locations. From there, its algorithms are able to calculate the rate at which methane is escaping into the atmosphere as well as the total emissions from both individual points or regions.
The key points about these data vendors is that they provide global data and are publicly available. Having more data sources like this would thus be advantageous to Contemporaries, as they help tackle the issue of geographic informational inequality and enforce the reliability of data due given the freedom of public scrutiny. However, there needs to be more of these types of data sources, and data vendors of this sort for the other two aspects of ESG, Social and Corporate Governance.
Additionally, corporate disclosures are also an important source of information that has not been given much attention by Contemporaries, although to avoid greenwashing or providing a false image of a company, mandates should be put in place to enforce responsible auditing on ESG issues and more transparency in company disclosure.
Perhaps a better approach for Contemporaries would be to look at the discrepancies between the corporate disclosures and the internet data they collect. Discrepancies could be used as a confidence weighting on ratings for each component of ESG to ultimately determine a company’s rating. Whatever approach is taken, it is clear more data and richer sources are therefore required to help tackle the issues of volatility and increasing equality in access to data.
What are the issues we should be worried about?
A major concern regarding Contemporaries is that they are run by profit-driven businesses. Their work is not publicly available or free to use, whereas Traditionalists like Sustainalytics have their ESG ratings available online. There is a lot of talk on how powerful their AI tools are, but not much of it can be publicly scrutinised unless you request for a demo, raising justifiable concerns about how reliable the technology actually is. Moreover, Contemporaries tend to be closely associated with international financial centres (IFCs)—TVL, for example, is owned by FactSet, an American financial services company in the second largest hedge fund state of Connecticut.
Since their methodology is not easily examined, due to limited disclosure, it is unclear as to the sort of parameters used or the assumptions made in developing its ESG rating model. Moreover, algorithms are not entirely subjective in their creation. They are subject to the scrutiny of their makers, and may have certain biases or other hints of subjectiveness built into the foundation of their AI system. Without a true value to compare ESG ratings with, we would not be entirely sure how well the algorithm really works.
With the uncertainty in the actual formulation of the Contemporary AI analyst, and the profit-driven nature of Contemporaries themselves, Hughes, Urban, and Wójcik (2021) note that Contemporaries may “complement rather than substitute” the Traditionalists in the near future (Hughes, Urban, and Wójcik, 2021). Although Contemporaries seem more democratic and provide more transparent ratings, Contemporaries also have areas of opaqueness that can affect the way ESG ratings are done. Perhaps if Contemporaries were not profit driven and rather more altruistic could AI truly replace traditional approaches to ESG.
Check out my first two posts on AIs and ESG benchmarking: ESG Benchmarking: Part #1 - ESG, AI, and the Investment Industry and ESG Benchmarking: Part #2 - AI, the New Hope of ESG
By MUHAMMED Hazim, Segi Enam intern, 30 Aug 2021 | LinkedIn
Edited by Nadirah SHARIF