In this blog, we explore how stocks have become more sensitive to environmental, social and governance (ESG) controversies breaking online.
- Research by MathWorks demonstrates that stocks with high ESG controversy scores exhibit greater downside returns and higher volatility. Therefore, avoiding controversial stocks can significantly enhance portfolio performance and Sharpe ratios.
- Stocks associated with a higher level of controversial online chatter underperformed in a study using the aggregate ESG controversy score, based on the study universe including 3,000 U.S. stocks with the most media buzz in the prior month.
- The Refinitiv® MarketPsych ESG Analytics (RM-ESG) dataset uses real-time natural language processing to capture these ESG-related discussions in the forums, to manage emerging risks before their impact is fully digested into stock prices.
For more data-driven insights in your Inbox, subscribe to the Refinitiv Perspectives weekly newsletter.
According to Lipper, global AUM for ESG funds as of February 2022 was $6.88 trillion – a 732 percent increase from $939 billion in January 2004.
With the emergence of ESG-minded online communities, there is a continuous flow of information on companies’ ESG performance and controversies.
Natural language processing can be used to generate ESG-focused data feeds based on news and social media chatter. Quantitative research has shown that stocks that are associated with a higher level of controversial ESG references underperform.
Research by MathWorks builds upon published research and demonstrates that momentum in ESG controversy scores impacts stock market returns.
The biggest challenge when testing such strategies is found in the diversity of ESG themes, and the large quantity of data. Whether this is unstructured or structured data, MATLAB offers a wide range of flexible tools to simplify this work and minimise the probability of introducing biases that can significantly affect the results.
Natural language processing with RM-ESG
The multi-dimensional Refinitiv® MarketPsych ESG Analytics (RM-ESG) dataset is used by quants to manage risk and forecast global stock prices.
A complex natural language processing (NLP) engine produces scores on environmental, social and governance-themed (ESG-themed) references to companies.
The underlying NLP engines scan and interpret the comments of tens of millions of authors in thousands of news and social media sources in real-time. Below is a simplified visual depiction of the three-stage process of extracting key themes from media text and aggregating them into scores.
The RM-ESG dataset publishes up to 45 controversy scores for over 100,000 global companies.
Rather than scoring sustainability from a company’s reports and press releases, the RM-ESG feed provides an external perspective on public perceptions of corporate activities. This external approach is especially suited to identify breaking bad news.
High media ESG controversy stocks have lower stock market returns
Research demonstrates that stocks with high ESG controversy scores exhibit greater downside returns and higher volatility (Teodoro, Clark-Bell, & Peterson, 2022). Avoiding controversial stocks can significantly enhance portfolio performance and Sharpe ratios.
The aggregated ESG controversies score is a general score that reflects the percentage of all ESG-related chatter about a company that concerns its association with ESG controversies.
Importantly, this score is a relative ranking within each industry.
Each industry has a similar number of companies with ESG controversy scores of 100 (lowest controversy) and 1 (highest controversy) constituents. This relative ranking prevents bias in favour of technology stocks (which tend to have fewer ESG controversies) or away from oil and gas stocks (which are associated with more controversies).
Stocks associated with a higher level of controversial online chatter underperformed in a study using the aggregate ESG controversy score. The study universe included 3,000 U.S. stocks with the most media buzz in the prior month, where buzz represents the total count of ESG-related media references over the prior period.
First, the average ESG controversies score was determined for each stock over the prior month. We then simulated investment returns over the following month in three hypothetical portfolios:
- The 20 percent of stocks with the lowest controversy scores (green line).
- The total basket of covered stocks (the dotted line).
- The 20 percent of stocks with the most controversies (the red line).
Each portfolio’s growth was plotted without considering transaction costs.
The U.S. stocks associated with more ESG controversies (red line) have underperformed since 2012. The spread between the highest and lowest controversies stocks was 2.3 percent annually. Additional testing statistics are detailed in the caption.
Independently from their higher share price returns, stocks with higher ESG controversies are also more volatile over the following month than stocks with fewer ESG controversies. This relationship appears since 2007 (see Teodoro et al, 2022).
Momentum in ESG controversy scores impacts stock market returns
Another signal tested during research conducted by MathWorks is ESG controversies momentum, i.e., the tendency for high controversy scores to persist over time.
When MathWorks compared portfolios sorted by ESG controversies momentum factors they observed a negative (positive) impact on equity portfolios from companies with higher increases (decreases) in controversies related to ESG topics.
These strategies were implemented by looking at:
- Companies with increasing controversies: these are companies that fall in the lower quintile of the ESG controversy momentum distribution.
- Companies with decreasing controversies: these are companies that fall in the higher quintile of the ESG controversy momentum distribution.
The ESG controversy momentum factor was obtained by:
- Computing the monthly buzz-weighted average of companies’ daily ESG controversies scores, where the buzz value refers to the number of media references to a given company related to ESG topics.
- Smoothing the series of monthly aggregated scores with a simple moving average.
- Computing the rate of change over two-months of the smoothed series.
Several back-tests were then performed with a monthly portfolio rotation on companies in the S&P500 and for multiple ESG-related issues separately, such as emissions controversies, management controversies, human rights controversies, etc., using MATLAB’s Backtest Framework.
Finally, they measured the annualised return difference between the bottom and top quintile portfolios.
As shown in Figure 1, of the 32 controversies tested, 22 showed out-performance of the decreasingly controversial over the increasingly controversial.
The best performing portfolios were those with decreasing community, management and credit controversies, while only for emissions topics did the bottom quintile portfolio underperform considerably from the top quintile.
On an aggregate level, portfolios built with governance controversies have performed better than those built with social and environmental controversies. Figure 2 shows the equity curves for one such governance controversy.
Impact on stock prices from news and forum chatter
Real-time news, and online investment social communities who amplify controversial news, appear to have an impact on stock prices.
With real-time natural language processing of the discussions in these forums, data feeds based on NLP can capture emerging risks before their impact is fully digested into stock prices.
Identifying companies’ trends in ESG controversies may provide valuable insight for mitigating risk in equity and fixed income portfolios.
Companies with an increasing trend in ESG controversies may potentially suffer from reputational damage and long-term investment outflows, whereas companies with decreasing controversies may benefit from institutional investors’ active engagement processes and gain greater allocations, due to improving ESG practices.
The biggest challenge when testing such strategies is found in the diversity of ESG themes, and the large quantity of data. Whether this is unstructured or structured data, MATLAB offers a wide range of flexible tools to simplify this work and minimise the probability of introducing biases that can significantly affect the results.
Discover more on MATLAB for Data Analysis: explore, model, and visualize data
References
Teodoro, T; Clark-Bell, J; Peterson, R. 2022. “The Impact of ESG Controversies on Stock Returns.” MarketPsych Data Whitepaper. Available upon request.
