Sustainable Investing Using a Data Driven Approach

Sustainable investing using a data driven approach

By Antoine Amend, Technical Director, Financial Services, Databricks

As the world moves towards a greener future, companies are setting themselves measurable sustainability targets, and more importantly holding themselves accountable for them. But how are they setting these targets, and what do they have in place to make sure they are reaching them?

Investors are beginning to re-evaluate the traditional portfolio approaches, compared to before when how sustainable a business was wouldn’t have been given a second thought. A recent study by Morgan Stanley, suggests that sustainable investing is now worth nearly $9 trillion worldwide and we are continuing to see this grow. The ‘millennial investor’ is seen to be driving this change with 86% of them having a strong interest in sustainable investments. Along with this new wave of investors, new regulations and cultural shifts, such as the net-zero emissions targets we are seeing across Europe, are also contributing towards a rise in sustainable investments.

It is without doubt that the future of finance goes hand in hand with social responsibility, and in order to stay competitive, financial services are going to have to disclose more information on their Environmental, Social and Governance performance (ESG). With a greater understanding of how sustainable an investment is, financial institutions will be able to mitigate reputational risk along with keeping an open channel of communication with their customers.

Improving ESG ratings are not only driven by customer demand and concerns over global warming, but also by economics. A better ESG rating is proven to have a positive correlation to profitability, whereas lower ESG ratings seem to correlate with volatility. But measuring the sustainability of an investment is no easy feat.  Without the ability to analyse data efficiently, financial institutions will open themselves up to the risk of investing in companies that aren’t aligned with their sustainability initiatives.  There is clearly a case for companies to improve on their sustainability, but this comes with a challenge: how is this measured?

The challenge

The current problem with measuring sustainability is that a lot of it seems to be guesswork; not clearly identifying companies that are simply ‘green washing’ versus those that are making tangible progress. Many different studies and research have been trying to define what it means to be sustainable. This has resulted in an abundance of Sustainability Reporting Tools.  Although useful, this traditional approach has its issues, mainly the lack of standardisation for criteria, as different countries have their own frameworks (and even within one country these frameworks often act as guidelines rather than a concrete set of standards). Investors and executives believe that by having one sustainability reporting standard, supported by legal mandates requiring companies to issue these reports, would not only increase the number of companies reporting but also improve on their accuracy.

As it stands, companies are choosing different types of frameworks to follow and with the scope and depth of their disclosures, sustainability results can differ completely.  However, despite the lack of uniformed reporting, we have now got a treasure trove of useful data, which given the right approach, will enable financial institutions to understand patterns and trends in companies’ responses to their sustainability initiatives.

A true measurement of sustainability 

An innovative approach to sustainable investing is through using natural language processing (NLP) techniques and graph analytics. This helps to extract key strategic ESG initiatives and to learn companies’ relationships in a global market and their impact on market risk calculations. NLP does not rely on existing taxonomy and as such, can be applied to learn key themes and policies. By using a unified data analytics platform, such as a lakehouse which allows for data versioning, governance, security and ACID properties that are needed even for unstructured data like video, audio and text, asset managers are able to assess how sustainable an investment is with a holistic and data-driven view of their ESG strategy.  This is done by using machine learning (ML) to extract key initiatives communicated by companies – for example, how they might be mitigating climate change, investing in communities or how they value their employees. The data is extracted and then scored against ESG metrics before being compared with actual media coverage from news analytics data, providing financial institutions with a better understanding on whether a company is achieving its ESG initiatives.

An example of using a unified data analytics platform, such as a lakehouse, is as follows. After successfully identifying and comparing businesses side by side across different ESG initiatives, it is important to distinguish an ESG score. However, it is important for the score not to be subjective but truly data-driven. In other words, you do not want to solely base your assumptions on companies’ official disclosures but rather on how companies’ reputations are perceived in the media, across all three environmental, social and governance variables. This is an important part of understanding whether a company is achieving its ESG goals.

What is really interesting, and something to consider, is that companies and businesses can be inter-connected, and the ESG performance of one (e.g. seller) may affect the reputation of another (e.g. buyer). As an example, if a firm keeps investing in companies directly or indirectly related to environmental issues, this risk should be quantified and must be reflected back on companies’ reports as part of their ethical investment strategy. An example of this is Barclays. In 2018 its reputation was heavily impacted due to indirect investments in a tar sands project.

In Summary 

It is clear that sustainable investing is here to stay with Bloomberg predicting that global ESG assets are on track to exceed $53 trillion by 2025. To have the ability to analyse complex documents and quickly summarise key ESG initiatives to better understand the sustainability aspect of investments at scale is of a different complexity and requires advanced use of data science. Without clear standardisation of how sustainability reporting should be defined globally, the only true answer to finding out a potential investment’s ESG rating is through a data-driven approach.