Issues Archive » FundTech Autumn 2019

ESG data: Standard practice

TechnologyThird-party ESG data services are plentiful but rarely aligned or transparent, often leaving fund managers frustrated. Nicholas Pratt asks if technology is the answer.

ESG investing may be one of the biggest growth areas in the funds industry, but there is a significant data problem at the heart of it.

While companies may only make limited disclosures on ESG matters, there are vast numbers of third-party providers offering ESG ratings and data services to investors and asset managers. At the end of 2016, more than 125 ESG data providers were in operation, according to the Global Initiative for Sustainability Ratings. These include both well-established, global data providers like MSCI, Bloomberg and FTSE; ESG specialists like Sustainalytics, Vigeo Eiris and TruValue Labs; and specialists within the ESG world such as Trucost, which focuses on carbon data.

However, there is very little correlation in what data they choose to collect, how they estimate any missing data, how they weight various ESG factors and, ultimately, the ratings they produce.

For fund managers and investors, it is a serious issue. As Rakhi Kumar, head of ESG investment and asset stewardship at State Street Global Advisors (SSGA), says: “Managers are allocating capital based on that data. It is driving investment decisions.”

Such are the limitations of this data, Kumar says that any managers using a third-party ESG data service will be following the views and beliefs of their providers and not their own. She would rather see companies mandated to report more of their own ESG credentials, rather than relying on third-party providers aggregating whatever data they can get their hands on and estimating whatever data they cannot.

“Why do we need third-party ESG data providers? Shouldn’t we get companies to provide their own ESG data and then build a ratings framework around this information in the same way we do with financial data?” says Kumar.

Consequently, SSGA has developed its own ESG scoring system, which it calls R-Factor. Nor is it alone. Dutch asset manager RobecoSAM has used its own Corporate Sustainability Assessment since 1999. Both firms see these proprietary systems as the best way to overcome the limitations and challenges of ESG data.

Intangible qualities
Lisa Beauvilian, head of sustainability and ESG at Impax Asset Management, sees ESG data as part of a wider strategy to include more ratings and scrutiny of companies’ so-called ‘intangible’ properties and ratings. “If we look at the S&P 500, 85% of the value comes from intangibles,” she says. “You can’t directly quantify those metrics like you can a balance sheet. The opportunity with ESG data is to look at all of those intangibles and understand the real value of a company.”

However, with every great opportunity comes a host of challenges and problems that need resolving. “The first problem is missing data,” says Beauvilian. “It is not just the environmental and social data but, fundamentally, the governance data. The reason is because none of it is universally mandated. And what data there is can have errors and inconsistencies. We get some raw data from the ESG research providers but their respective methodologies can be hard to understand.”

If you’re a passive investor with indices covering thousands of companies, but with an ESG overlay, the fact that you can’t drill down into the methodology is a problem, says Beauvilain. “It is getting better and you can alter some of them to the way you want, but it is not always possible.”

Global standards
Efforts are underway to address this by establishing industry frameworks for ESG data, such as the Sustainability Accounting Standards Board (SASB) Conceptual Framework, which was finalised in November 2018 after a long development phase. “It is gaining a lot of traction because it sets a minimum standard for materiality and is designed to be updated as risks change,” says SSGA’s Kumar.

SSGA has adopted the SASB framework as part of its own R-factor scoring system, which takes information from three different data providers. Kumar would like to see more companies self-reporting their ESG data, but this requires some infrastructure and systems. “ESG reporting is not as developed as financial reporting. With our current systems, companies can produce quarterly sales figures but not always their emissions or employee turnover,” she says.

In the absence of that information, there are the third-party data providers looking to fill in the gaps, especially for the small and mid-cap companies without the resources to hire a sustainability head, says Kumar.

Awareness of data quality issues is growing, she says. “Single sources of third-party data are not the way forward. We have developed our scoring system to use multiple sources of data, but also to remove a lot of the noise around ESG data. However, the overall objective is to get companies to report this data directly to us and to have ESG data become part of mainstream investment research and analysis with the same lifecycle as financial risk data.”

The SASB framework is also key to the materiality debate, says Beauvilain. Companies are being bombarded with different reporting requests and there is little direction on what areas they should focus on. The SASB initiative is an important starting point in two ways, she adds. “It looks at ESG reporting from a sub-sector perspective; and it looks to widen the scope of ESG research from being shareholder-driven to stakeholder-driven, including employees and customers as well as investors and shareholders.”

Alongside the absence of standards and framework, there is also the need for more technology. One missing link, technology-wise, is location-based data for physical climate risk, says Beauvilain. “We have some climate models but we need to know where the companies’ assets are. It is crucial for a prudent investor to know this. We know where they sell into but not where they are manufacturing.

“That is where AI could help. I also think AI and technology could help in trawling through big amounts of data, but in many cases we need more information from companies and more standardisation. There are limits to what technology can manage.”

Initial trials with still-developing technologies such as machine learning, AI and natural language processing have yielded positive potential for ESG research and data, while also uncovering a number of inherit problems with these data sets, says Manjit Jus, head of ESG ratings at RobecoSAM. “The amount of data available is huge, and when corporate self-disclosure is supplemented by external data sources such as media coverage, social media and satellite imagery, powerful, new sets of ESG data can be created.

“However, it also highlights the inconsistency of this data, quality issues and the strong need to have human analysis play a deciding factor before data sets can flow into ESG ratings or investment processes. This requires analyst expertise both to understand the fine print and for additional data cleaning.”

That said, the potential to extract large sets of information from publicly available sources and to form ESG risk and opportunity profiles for companies is becoming much easier through new forms of technology, says Jus.

“On the corporate side, technologies that support digital data dissemination in a cost-efficient and more real-time manner could play an important role in getting high-quality data to investors in the near future. Technologies like blockchain are already being tested in the areas of corporate supply chains to improve transparency and traceability of products, services and ultimately data sets being collected.”

The third parties
The third-party data providers recognise that difficulties in the market exist. “The first challenge is that there is no regulatory body dictating how companies report their ESG data,” says Hendrik Bartel, co-founder and CEO of TruValue Labs. “Instead we have lots of different third-party providers employing different frameworks. Some of them measure policies, others attribute consumption. There is very little correlation between them all.”

This is why you need technology, says Bartel. “TruValue uses AI technology to read hundreds and thousands of unstructured ESG data sources. It is an uber-analyst. It can quantify, structure and understand (cognitively and semantically) all of this data and attribute it accordingly. It would take a normal analyst six years to go through the information we get in one day just on the automotive industry.”

One criticism of both ESG data providers and AI technology is a lack of transparency – the methodology behind the ratings or the inner workings of the AI. “We are transparent about the methodology and you can follow the raw data that is used and what is labelled as material by the AI,” says Bartel. “But we do not disclose any coding.”

Does there need to be greater regulatory involvement in the ESG data and ratings market? “It will take another five to eight years until we have the regulatory body in place,” says Bartel. “We have more companies listening to investors about ESG so there is both supply and demand. We just need a regulator to make it more efficient. I think there will be different frameworks in different regions rather than a one-size-fits-all – so one for the US, Europe, China and Japan.”

Another challenge with ESG data is that it relates to sustainability in the broad sense and is therefore multi-source, multi-dimensional and multi-purpose, says Fouad Benseddik, head of methods and institutional affairs at Vigeo Eiris. “In addition to indicators and narratives from issuers, asset managers must consider and interpret information and opinions from relevant stakeholders such as NGOs, trade unions, consumers, the media and expert analyses.”

Ratings and opinions from the research of sustainability rating agencies such as Vigeo Eiris are crucial in this regard, says Benseddik. “ESG data, when it is well analysed, serves the acuity and extends the range and diversity of investment practices and decisions, ranging from simple screening on exclusion criteria for products such as tobacco, alcohol or involvement in animal testing to the most innovative and elaborate analyses for the implementation of structured investment vehicles such green and social bonds, indices and so on.”

The rights of society
As with any market, transparency is essential not only for the intelligibility of products but also for the trust that is crucial, says Benseddik. “The challenge of ESG data is not just a question of transparency or sincerity, because no one can durably play with the truth of information. The challenge lies more in the relevance, accuracy and completeness of the data and in the materiality of the topics and risks that the data reflects.

“Transparent and sincere but irrelevant information is not enough. There is a need for ESG data that informs the willingness and ability of issuers to report on their levels of engagement and the impacts of their products and their behavior with regard to the expectations, interests and rights of society and the environment.”

The biggest problem with ESG data is its intangible and qualitative nature and the unstructured environment it comes from, especially when compared to financial data. For example, there is a broad correlation between the different credit rating agencies because the metric they are scoring is well defined and very closely related, says Bob Mann, president of Sustainalytics.

However, in the ESG world, there are different frameworks for different intentions. “The question for investors is what metrics are they looking to assess and is the rating they are purchasing designed to measure what they are looking for?” says Mann.

ESG boxWhile some ESG rating providers are relatively clear on what they are measuring, others are less so, he adds. “We offer a great deal of transparency on our ratings and how every metric is weighted. But for some of the new players using smart technology and AI, there is a black box at the heart of it and that makes it difficult to be transparent.”

Mann says that Sustainalytics has used machine learning algorithms and AI to find trends among the data. “We were able to build smart engines that looked at our databases and market data and found correlations. However, they were not able to explain why those correlations existed.”

He supports standardisation in terms of generating material ESG information but says that the interpretation of that data should remain open. “Not every equity analyst uses the same framework to generate their buy/sell decisions. I would not want the SASB framework, for example, to be too rigid and to rob the market of important IP.”

Mann believes that smart technologies are an important tool that can be used to increase the insight generated by ESG ratings. However, he is also wary of too great a reliance on technology to generate ESG ratings. “The assessment of any information should be a combination of machine-driven processes and human curation. The question is how much of each,” he says.

“We use technology quite intensively and we have a smart technology team to scan a tremendous array of information, to look at historical changes in reporting and to help analysts make predictions. But the analysts remain the final curators of that information.”

©2019 funds europe