BIG Data: Getting Granular with ESG Factors

By Ivy Schmerken, Editorial Director

With the growth in sustainable investing, there’s been a surge in data on environmental, social and governance (ESG) factors over the past few years.

Demand for ESG data is rising as asset managers look to incorporate ESG factors such as low-carbon emissions or gender diversity on boards into their investment analysis and decision-making processes.

Fund managers, including BlackRock and Vanguard, are offering sustainable funds and exchange-traded funds (ETFs) based on sustainable indexes to capture assets from millennials and women.

But the uptake has moved beyond specialty funds and has spread to pension funds, particularly in Europe, looking for long-term returns, reported Bloomberg Intelligence in April.

“The financial cost of environmental, social and governance (ESG) performance and better disclosure is spurring uptake,” wrote Bloomberg Intelligence in “Sustainable Investing Grows on Pensions, Millennials.”  Issues pertaining to energy efficiency, water scarcity, safety, and diversity are gaining wider investor interest, even without the sustainability label, Bloomberg wrote.

An estimated 26% of assets measured globally — or almost $23 trillion — falls under the field of sustainable or socially responsible or ethical investing, according to Bloomberg.

Pension funds in Europe are allocating to ESG managers and some regard this as part of their fiduciary responsibility.

Some are reacting to stricter regulations on greenhouse gas emissions or shareholder resolutions on executive compensation policies.

“The idea that investors who integrate corporate environmental, social and governance risks can improve returns is now rapidly spreading across capital markets on all continents,” wrote Georg Kell, founding director of the UN Global Compact, in a Forbes column“ The Remarkable Rise of ESG.  Kell is the founding director of the United Nations Global Compact, an organization founded in 2004 to promote integration of ESG issues in investment decision making.

Traditional market data and index companies such as MSCI, Bloomberg, and Refinitiv (formerly Thomson Reuters), have become major suppliers of ESG metrics and scores on public companies.  In July 2017, Morningstar, the independent research firm, acquired a 40% stake in Sustainalytics, a leading provider of ESG research and ratings.

While there is more ESG and sustainability information disclosed publicly there have been questions about which information is most useful or relevant to making financial decisions.

“Another major barrier has been a lack of data and the necessary tools to get a handle on the fragmented and incomplete information available,” wrote Kell, who is currently chairman of Arabesque, a London-based ESG quant fund manager using big data to quantify the performance of global public companies on ESG factors.

A Role for AI?

Eyeing the demand for more accurate data and metrics, some fintech startups contend that A.I. has a role to play in ferreting out ESG issues that could have a material impact on a company’s financial performance.

“Traditionally, ESG data is sourced out of a company’s self-provided data and is then looked at by analysts through their subjective lens,” said Hendrik Bartel, co-founder and CEO of TruValue Labs, a fintech startup that is applying artificial intelligence techniques including machine learning, deep learning, and natural processing to screen big data sets for ESG issues on public companies.

When the TruValue entered the business five or six years ago, “there was very little correlation between ESG data providers,” said Bartel, who presented at the AIR Summit in September.  “I wanted to build something objective, to be scalable and very transparent and provide performance in real time,” said Bartel.

Traditional ESG ratings are created by companies such as MSCI and Sustainalytics that focus on an analyst-based ESG ratings model, which are based on what corporations share with these ratings firms, asserts TruValue Labs in an emailed response to a question.  However, TruValue Labs contends there is the potential for analyst bias in the analysis and that the scores are not revisited on a frequent enough basis, sometimes yearly.

“What makes TruValue Lab’s approach different is that it’s sourcing up-to-the-minute information, with objective application to all companies using SASB’s materiality framework,” wrote a spokesperson in an email referring to the Sustainability Accounting Standards Board.

TruValue’s Insights 360 Engine integrates the SASB’s Materiality Map, which identifies sustainability issues that are likely to affect the financial condition or operating performance of companies within an industry.

On a daily basis, TruValue Labs scans tens of thousands of unstructured data sources on the Internet and composes a score for each of the SASB materiality factors on over 8,000 companies over a 10-year history. The company taps into information from industry-specific publications, non-company reported regulatory filings, news reports, government agency studies, trade blogs, ESG thought leader-shared Twitter articles and reports from watchdog groups and NGO [non-governmental] organizations.

It uses natural language processing (NLP) software to read articles, categorize items and glean positive and negative sentiments from the writing, which allows it to produce a number of possible predictive indicators. From a risk or compliance standpoint, investors can dig into 30 categories of underlying data to look at how certain ESG factors such as waste or hazardous materials management are negatively impacting their portfolios. They can also roll up all the ESG scores at the portfolio level to see how they are performing on a specific ESG issue, said Bartel.

Quant ESG Strategies Emerge

Now, the availability of these big data driven-ESG signals is making it possible to develop quantitative strategies. “AI is enabling investors to access massive amounts of information from unbiased sources, with more frequency, granularity and real-time analysis,” wrote ThinkAdvisor, citing a study by Cerulli Associates in “AI is Helping Open ESG Investments to Quant Funds.”

According to Cerulli, up until now quant funds were locked out of the sustainable investing space because weaknesses in the data prevented them from building strategies.

Despite some of the weaknesses in company-published ESG data, big data and artificial intelligence have enabled the construction of new data sets for analyzing investments.

For example, London-based Arabesque is using self-learning quantitative models and big data to assess the performance and sustainability of globally listed companies.

The firm’s algorithmic tool systematically combines over 200 ESG metrics with news signals from over 50,000 sources across 15 languages.

On Oct. 2^nd, Arabesque said it agreed to provide the Swedish national pension fund Första AP-fonden (AP1) with ESG data through its proprietary algorithmic-based technology Arabesque S-Ray, according to an announcement.

As part of their partnership, both parties will collaborate on developing a unique score for S-Ray that will assess the compliance of thousands of companies with the UN Guiding Principles on Business and Human Rights (UNGP Score), states the release.

In addition, the company offers Arabesque Prime, a global equity fund that combines sustainability information with a sophisticated fundamental screening process.

In TruValue Lab’s case, the firm’s chief data scientist has been able to develop multiple strategies for the S&P 500 and the Russell 1000 index, which Bartel said has added 4-5% in returns annually. Whereas most fund companies are using traditional data sets such as the MSCI USA and FTSE 100 to design strategies that typically track a benchmark, TruValue Labs’ strategies are outperforming the market,” said Bartel at AIR Summit.

A new study by Harvard Business School Professor George Serafeim demonstrates that significant positive alpha can be extracted by using a combination of TruValue Lab’s big data and traditional ESG ratings from MSCI, according to an Oct. 24 press release posted on the startup’s web site.

Serafeim’s study, “Public Sentiment and the Price of Corporate Sustainability,” analyzes data for the years 2009-2018 provided by TruValue Labs and by MSCI. Serafeim used MSCI ratings for ESG performance due to their industry-wide prevalence and employed TruValue Labs’ data as a source to find sentiment in semantic big data for ESG topics.

Sustainability Trend Here to Stay

Despite the mixed state of company-published ESG data, Bloomberg Intelligence suggests that this is creating opportunities similar to the early days of emerging markets or other new fields where the data was nonstandard and uneven disclosure.

Investor interest in ESG funds, alongside market appreciation, drove a 37% annual increase in assets to $445 billion in 2017, according to Bloomberg.  Meanwhile, the growth in numbers of ESG funds is being fueled by the rise of ETFs. ESG ETFs had inflows of $1.3 billion in the first two months of 2018 vs. $3.2 billion for all of 2017, said Bloomberg, suggesting that growth is gaining momentum.

Portfolio managers and analysts are increasingly incorporating ESG factors into their investment analyses and processes, according to a 2018  report by the CFA Institute and the UN’s Principles for Responsible Investment (PRI).  The report “ESG Integration in The Americas: Markets, Practices, and Data,” is based on surveying 1,100 financial professionals, mainly CFA members around the world. However, it finds that ESG integration remains in its relative infancy with investors and analysts calling for more guidance on exactly how they can do ESG and integrate data into their analysis.

Investors acknowledge that ESG data has come a long way, but advances in quality and comparability of data still have a long way to go, states the report.

Though skeptics may argue that responsible investing is a fad, Arabesque’s Kell argues that it’s not going away.

“Technology and the rise of transparency is here to stay,” wrote Kell. “Gathering and processing data will become easier and cheaper. Smart algorithms will increasingly allow for better interpretation of non-traditional financial information which seems to be doubling in volume every few years,” wrote Kell.

Past blog posts related to Data/Artificial Intelligence/Machine Learning Issues

AI on the Fast Track

Data Science Platforms Help the Buy-Side Integrate Alternative Data

Algo Development 2.0 Looks to Open Source, Cloud & Big Data

Artificial Intelligence Gains Momentum

Alt Date on the March

Augmented Reality Becomes a Reality

Buy-Side Delves into Mobile Data